Sémantická interoperabilita v biomedicíně a zdravotnictví
Editoři: Štěpán Svačina a Jana Zvárová Podpořeno projektem SVV-2015-260158
Fotografie na titulní straně: Karel Meister Grafický návrh: Anna Schlenker
© 2015, Autoři uvedení v textu. Všechna práva vyhrazena. Žádná část sborníku nesmí být kopírována či rozmnožována za účelem dalšího rozšiřování v jakékoliv formě či jakýmkoliv způsobem, ať již mechanickým nebo elektronickým, včetně pořizování fotokopií, nahrávek, informačních databází, bez písemného souhlasu vlastníka autorských práv. Korektury: Mgr. Růžena Písková, Ph.D.
S´ emantick´ a interoperabilita v biomedic´ınˇ e a zdravotnictv´ı ˇ ep´ Prof. MUDr. Stˇ an Svaˇ cina, DrSc., MBA Prof. RNDr. Jana Zv´ arov´ a, DrSc.
Semin´ aˇre o s´emantick´e interoperabilitˇe v biomedic´ınˇe a zdravotnictv´ı se staly jiˇz pravideln´ ym podzimn´ım f´orem, kde prezentuj´ı sv´e pr´ ace zejm´ena studenti doktorsk´eho studijn´ıho programu Biomedic´ınsk´ a informatika. Biomedic´ınsk´ a informatika je jedn´ım z nejmladˇs´ıch obor˚ u doktorsk´eho studia biomedic´ıny, kter´e bylo novˇe koncipov´ ano od 90. let a zahrnuje dnes t´emˇeˇr dvˇe des´ıtky obor˚ u. T´emata publikovan´ a ve sborn´ıku prac´ı ukazuj´ı, ˇze biomedic´ınsk´ a informatika se stala stejnˇe v´ yznamn´ ym oborem, jako jsou klasick´e medic´ınsk´e obory doktorsk´eho studia, tedy jako je tˇreba fyziologie, biofyzika nebo neurovˇedy. Obor biomedic´ınsk´e informatiky je dnes tak´e oborem pro habilitaˇcn´ı a jmenovac´ı ˇr´ızen´ı. Biomedic´ınsk´a informatika – obor rozv´ıjej´ıc´ı se postupnˇe od 50. let minul´eho stolet´ı – se tak dnes stal plnohodnotnou souˇc´ast´ı medic´ıny teoretick´e, klinick´e i l´ekaˇrsk´e vˇedy. Pod´ıv´ame-li se zpˇet na ˇsest roˇcn´ık˚ u semin´ aˇre, zaznamen´ av´ ame vzestupnou u ´roveˇ n prac´ı a je dobˇre, ˇze jsou publikov´any ˇcesky i anglicky, jsou tak pˇr´ıstupn´e ˇsirˇs´ımu f´ oru zdravotnick´ ych pracovn´ık˚ u. S´emantick´ a interoperabilita a schopnost z´ıskat konkr´etn´ı informace pˇri pouˇzit´ı technick´ ych prostˇredk˚ u jsou z´ akladn´ımi podm´ınkami pro vyuˇz´ıv´ an´ı technologi´ı telemedic´ıny v elektronick´em zdravotnictv´ı. Schopnost syst´em˚ u porozumˇet pˇren´ aˇsen´ ym u ´daj˚ um (s´emantick´a interoperabilita) vyˇzaduje pouˇzit´ı stejn´e terminologie (tj. klasifikaˇcn´ıch syst´em˚ u a ˇc´ıseln´ık˚ u) pˇri pouˇzit´ı stejn´eho jazyka pro komunikaci a jeho z´ aznam (datov´e standardy). Pokud se informace v biomedic´ınˇe a zdravotnictv´ı sd´ılej´ı pomoc´ı voln´eho textu, pˇredpokladem pro s´emantickou interoperabilitu je zjiˇstˇen´ı jeho v´ yznamu. Pravdˇepodobnˇe nejlepˇs´ı aplikovateln´ y obecn´ y klasifikaˇcn´ı syst´em pro zdravotnictv´ı je SNOMED CT. Tento syst´em vznikl spojen´ım americk´eho syst´emu SNOMED (vytvoˇren´ y Asociac´ı americk´ ych patolog˚ u) a Britsk´e klinick´e terminologie
( Read codes“). V souvislosti s touto f´ uz´ı byla v roce ” 2007 ustavena International Health Terminology Development Organization (IHTSDO) se s´ıdlem v D´ ansku. IHTSDO je neziskov´e sdruˇzen´ı, kter´e vyv´ıj´ı a propaguje pouˇz´ıv´an´ı SNOMED CT pro podporu bezpeˇcn´e a u ´ˇcinn´e v´ ymˇeny informac´ı ve zdravotnictv´ı. SNOMED CT je klinick´a terminologie, kter´a je povaˇzov´ana za nejkomplexnˇejˇs´ı v´ıcejazyˇcnou terminologii ve zdravotnictv´ı na svˇetˇe. SNOMED CT je v souˇcasn´e dobˇe pouˇz´ıv´ an v cel´e ˇradˇe informaˇcn´ıch syst´em˚ u pro zaznamen´av´an´ı klinick´ ych informac´ı do z´aznam˚ u o pacientech.
Obr´ azek 1: Faust˚ uv d˚ um.
Ve sborn´ıku semin´aˇre S´emantick´a interoperabilita ” v biomedic´ınˇe a zdravotnictv´ı“, konan´em 17. z´ aˇr´ı 2015 v Akademick´em klubu 1. l´ekaˇrsk´e fakulty UK ve Faustovˇe domˇe je publikov´ano jednadvacet kr´atk´ ych origin´aln´ıch pˇr´ıspˇevk˚ u doktorand˚ u v ˇceˇstinˇe a angliˇctinˇe. Sborn´ık tak tvoˇr´ı pˇrehledn´ y doplˇ nuj´ıc´ı materi´al k vlastn´ım pˇredn´aˇsk´am doktorand˚ u na semin´aˇri.
Část I – Česky Sémantická interoperabilita v biomedicíně a zdravotnictví
Part I – Czech
1
Obsah
3–5
Management n´ asledn´ych vyˇsetˇren´ı ˇst´ıtn´e ˇzl´ azy v tˇehotenstv´ı Bart´ akov´ a J., Jiskra J.
6–9
Bezpeˇcnost osobn´ıch dat a Big Data v biomedic´ınˇe Berger J., Beyr K.
10–12
Role jednonukleotidov´ych polymorfism˚ u v patogenezi vysokostupˇ nov´ych gliom˚ u Bielnikov´ a H., Buzrla P., Bielnik O., Tomanov´ a R., Urbanovsk´ a I., Hruˇskov´ a L., Dvoˇr´ aˇckov´ aJ., Mazura I.
13–14
O vlastnostech genov´e konverze Gergelits V.
15–17
Detekce mutac´ı v genech kolagenu typ I Hruˇskov´ a L.
18–27
Statistick´e metody pro tvorbu ˇcasovˇe z´ avisl´ych percentilov´ych graf˚ u pro hodnocen´ı velikosti plodu a dataci gravidity na z´ akladˇe longitudin´ aln´ıch dat Hynek M., Long J.D., Stejskal D., Zv´ arov´ a J.
28–31
V´yznam deficitu cerebr´ aln´ıho fol´ atu pro v´yvoj a progresi autismu ˇarek M. Krsiˇcka D., S´
32–33
Integrace zaˇr´ızen´ı pro l´eˇcbu diabetu a lifestyle zaˇr´ızen´ı v r´ amci mobiln´ı aplikace pro self-management diabetu Muˇzn´y M., Vlas´ akov´ a M., Muˇz´ık J., Arsand E.
34–35
Porovn´ av´ an´ı internetov´ych str´ anek a l´ekaˇrsk´ych doporuˇcen´ych postup˚ u s vyuˇzit´ım datab´ az´ı ˇr´ızen´ych medic´ınsk´ych slovn´ık˚ u Rak D., Sv´ atek V.
36–38
Mezin´ arodn´ı komunikaˇcn´ı standardy a interoperabilita syst´em˚ u v ˇcesk´em zdravotnictv´ı Seidl L., Hanzl´ıˇcek P.
39–41
Velk´ a data v nemocniˇcn´ıch informaˇcn´ıch syst´emech z pohledu bezpeˇcnosti Schlenker A., Reimer M.
42–44
Sekund´ arn´ı katarakta u pacient˚ u po implantaci multifok´ aln´ıch nitrooˇcn´ıch ˇcoˇcek Siˇcov´ a K., V´yborn´y P., Paˇsta J.
45–46
DNA v biomedic´ınsk´ych aplikac´ıch Slov´ ak D., Zv´ arov´ a J.
47–50
Nestrukturovan´ a data ve zdravotnictv´ı zaloˇzen´em na d˚ ukazech Stonov´ a M.
51–53
V´yvoj pˇrevodn´ıho syst´emu u myˇsi ˇ nkov´ Saˇ a B., Beneˇs J., Sedmera D.
54–55
V´yskyt mikrobi´ aln´ı fl´ ory nosn´ıch pr˚ uchod˚ u a jej´ı vliv na rozvoj chronick´ych rhinosinusitid ˇ Steffl M., Plz´ ak J.
56–58
L´eˇcba glaukomu za 1 Kˇc dennˇe – sen nebo realita? Vesel´ a Fl´ orov´ a Z., V´yborn´y P., Siˇc´ akov´ a S., Obenberger J.
59–61
Telemonitoring z´ akladn´ıch terapeutick´ych prvk˚ u l´eˇcby diabetes mellitus a moˇznosti jejich hodnocen´ı Vlas´ akov´ a M., Muˇzn´y M., Muˇz´ık J.
62–64
V´yhody vyuˇzit´ı virtu´ aln´ıho pacienta v nel´ekaˇrsk´ych zdravotnick´ych oborech Vondruˇskov´ a L.
65–66
Pˇredzpracov´ an´ı l´ekaˇrsk´ych zpr´ av pro extrakci informac´ı Zv´ ara K., Tomeˇckov´ a M., Sv´ atek V., Zv´ arov´ a J.
67–68
Stroma dlaˇzdicobunˇeˇcn´ych n´ ador˚ u hlavy a krku ˇ Zivicov´ a V., F´ık Z., Dvoˇr´ ankov´ a B., Smetana Jr. K.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Bart´akov´a J., Jiskra J. – Management n´asledn´ych vyˇsetˇren´ı ˇst´ıtn´e ˇz´azy v tˇehotenstv´ı
Management n´ asledn´ ych vyˇsetˇren´ı ˇst´ıtn´ eˇ zl´ azy v tˇ ehotenstv´ı Jana Bart´ akov´ a1,2 , Jan Jiskra1 1 2
´ ˇ a republika Ustav biofyziky a informatiky, 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova v Praze, Praha, Cesk´
ˇ a republika 3. intern´ı klinika, Vˇseobecn´ a fakultn´ı nemocnice a 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova v Praze, Praha, Cesk´
Kontakt: Jana Bart´ akov´ a ´ Ustav biofyziky a informatiky 1.LF UK v Praze Adresa: Salmovsk´ a 478/1, 128 00, Praha 2 E–mail:
[email protected]
C´ıle v´ yzkumu
l´eˇcbou a modifikovateln´ ymi faktory v managementu hypotyre´ozy. Z 821 respondent˚ u by 67,7% kontrolovalo laboratorn´ ı tyreoid´ a ln´ ı testy v pr˚ ubˇehu tˇehotenstv´ı kaˇzd´e Naˇs´ım hlavn´ım c´ılem je zjistit, zda kontroln´ı vyˇsetˇren´ı ˇ c tyˇ r i t´ y dny, 21,4% kaˇ z d´ y ch osm t´ ydn˚ u, 7,9% kaˇzd´ ych 12 ˇst´ıtn´e ˇzl´azy u tˇehotn´ ych ˇzen l´eˇcen´ ych levothyroxinem ˇ t´ y dn˚ u a 2,9% kaˇ z d´ e dva t´ y dny (12). (LT4) jsou v Cesk´e republice v souladu s doporuˇcen´ım v souˇcasnosti platn´e smˇernice Americk´e Tyreologick´e AsoV naˇs´ı studii z let 2004–2014 bylo vyˇsetˇreno 188 ciace (ATA) 2011 nebo Endokrinn´ı Spoleˇcnosti (ES) 2012. tˇehotn´ ych ˇzen l´eˇcen´ ych LT4, kter´e byly n´aslednˇe sleNaˇs´ım druhotn´ y c´ılem je vyhodnocen´ı moˇznosti sn´ıˇzit dov´any a vyˇsetˇrov´any v souladu s doporuˇcen´ım smˇernice n´aklady modifikac´ı pouˇz´ıvan´ ych laboratorn´ıch test˚ u ˇst´ıtn´e ATA 2011. Vyˇsetˇren´ı ˇst´ıtn´e ˇzl´azy ve tˇret´ım trimestru (26.– ˇzl´azy v klinick´e praxi. 32. gestaˇcn´ı t´ yden) bylo vyhodnoceno v pˇr´ıpadˇe nekomplikovan´e hypotyre´ozy a/nebo pozitivn´ıch TPOAb jako nadbyteˇcn´e a produkuj´ıc´ı neadekv´atnˇe vysok´e n´aklady oproti Souˇ casn´ y stav pozn´ an´ı n´ızk´e pravdˇepodobnosti pˇr´ınos˚ u. Okolo 10–15 % tˇehotn´ ych ˇzen m´ a pozitivn´ı protil´atky proti tyreoid´aln´ı peroxid´ aze (TPOAb) [1, 2, 3] a aˇz 5 % m´a zv´ yˇsen´ y tyreoid´ aln´ı stimulaˇcn´ı hormon (TSH) [4, 5]. Negativn´ı dopad nel´eˇcen´eho onemocnˇen´ı ˇst´ıtn´e ˇzl´azy na plodnost, pr˚ ubˇeh tˇehotenstv´ı, poporodn´ı obdob´ı a v´ yvoj plodu byl jiˇz dobˇre pops´ an [5, 6, 7, 8, 9]. Dosud vˇsak nebyla publikov´ana ˇz´ adn´ a studie zamˇeˇruj´ıc´ı se na management n´asledn´ ych vyˇsetˇren´ı pˇri onemocnˇen´ı ˇst´ıtn´e ˇzl´azy v tˇehotenstv´ı v souˇcasn´e klinick´e praxi. V letech 2011 a 2012 byly publikov´ any smˇernice managementu onemocnˇen´ı ˇst´ıtn´e ˇzl´ azy v tˇehotenstv´ı dvou spoleˇcnost´ı – ATA a ES [10, 11]. Obˇe smˇernice obsahuj´ı doporuˇcen´ı t´ ykaj´ıc´ı se zah´ ajen´ı l´eˇcby LT4 a ˇcetnost n´asledn´ ych kontrol ˇst´ıtn´e ˇzl´ azy v pˇr´ıpadˇe, je-li l´eˇcba zapoˇcata. Mezi smˇernicemi existuje vysok´ y stupeˇ n shody: ˇst´ıtn´a ˇzl´aza by mˇela b´ yt testov´ ana pomoc´ı TSH kaˇzd´ y 4.–6. t´ yden v pr˚ ubˇehu prvn´ıho trimestru tˇehotenstv´ı. V druh´em a tˇret´ım trimestru smˇernice od ES doporuˇcuje pokraˇcovat v testov´ an´ı kaˇzd´ y 4.–6. t´ yden, zat´ımco ATA doporuˇcuje prov´est kontrolu TSH bˇehem tohoto obdob´ı pouze dvakr´at (jedenkr´ at ve druh´em trimestru a jedenkr´at ve tˇret´ım trimestru). Kromˇe m´ırnˇe odliˇsn´eho doporuˇcen´ı smˇernic se vˇsak m˚ uˇze liˇsit i samotn´ a endokrinologick´ a praxe. V roce 2013 byli poˇz´ad´ani kliniˇct´ı ˇclenov´e ES, ATA a AACE (Americk´a Asociace Klinick´ ych Endokrinolog˚ u), aby vyplnili webov´ y dotazn´ık obsahuj´ıc´ı 30 ot´ azek zab´ yvaj´ıc´ıch se testov´an´ım, S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Pˇrestoˇze se management onemocnˇen´ı ˇst´ıtn´e ˇzl´azy v tˇehotenstv´ı nejev´ı v principu komplikovanˇe, v klinick´e praxi tomu m˚ uˇze b´ yt jinak. Bylo by proto vhodn´e, aby vyuˇz´ıv´an´ı laboratorn´ıch tyreoid´aln´ıch test˚ u v pr˚ ubˇehu tˇehotenstv´ı byla vˇenov´ana vˇetˇs´ı pozornost.
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı Nevhodn´e vyuˇz´ıv´an´ı laboratorn´ıch test˚ u je v syst´emu zdravotnictv´ı ˇsiroce rozˇs´ıˇrenou a n´akladnou z´aleˇzitost´ı [13, 14, 15, 16]. Management onemocnˇen´ı ˇst´ıtn´e ˇzl´azy u tˇehotn´ ych ˇzen lze snadno modifikovat efektivnˇejˇs´ım vyuˇz´ıv´an´ım tyreoid´aln´ıch test˚ u, a to napˇr´ıklad na z´akladˇe studie doporuˇcovan´e ˇcetnosti kontroln´ıch vyˇsetˇren´ı ˇst´ıtn´e ˇzl´azy, studie skuteˇcn´e klinick´e praxe opakovan´eho testov´an´ı, pˇr´ıpadnˇe studie skuteˇcnˇe vyˇsetˇrovan´ ych hormon˚ u v endokrinologick´e praxi.
Podˇ ekov´ an´ı Tato pr´ace byla podpoˇrena projektem SVV-2015260158 Univerzity Karlovy v Praze.
3
4
Bart´akov´a J., Jiskra J. – Management n´asledn´ych vyˇsetˇren´ı ˇst´ıtn´e ˇz´azy v tˇehotenstv´ı
Kl´ıˇ cov´ a slova Onemocnˇ en´ı ˇst´ıtn´ eˇ zl´ azy
Zdroj: Clinique health [online]. 2015 [cit. 2015-2306]. Dostupn´e z: http://www.cliquehealth.org/ index.php?option=com_zoo&task=item&item_ id=1505&Itemid=4
Definition: Benign´ı nebo malign´ı stav ovlivˇ nuj´ıc´ı strukSNOMED CT: 103184005 turu ˇci funkci ˇst´ıtn´e ˇzl´ azy. MeSH: nenalezeno Synonyma: Porucha ˇst´ıtn´e ˇzl´ azy ICD10: Y42.8 Zdroj: The Free Dictionary [online]. 2015 [cit. 2015-2306]. Dostupn´e z: https://medical-dictionary. Levothyroxin thefreedictionary.com/
MeSH: C19.874
Definice: Hormon ˇst´ıtn´e ˇzl´azy pouˇz´ıvan´ y v podobˇe sodn´e soli k l´eˇcbˇe hypotyre´ozy, l´eˇcbˇe a prevenci strumy a karcinomu ˇst´ıtn´e ˇzl´azy.
ICD10: E00-E07
Synonyma: L-Thyroxin
Tˇ ehotenstv´ı
Zdroj: The Free Dictionary [online]. 2015 [cit. 2015-2306]. Available from: https://medical-dictionary. thefreedictionary.com/
SNOMED CT: 14304000
Definice: Obdob´ı od poˇcet´ı do porodu. Synonyma: Gravidita
SNOMED CT: 126202002
MeSH: D06.472.931.812; D12.125.072.050.767 Zdroj: The Free Dictionary [online]. 2015 [cit. 2015-2306]. Available from: https://medical-dictionary. ICD10: Y42.1 thefreedictionary.com/ SNOMED CT: 289908002 MeSH: G08.686.785.769 ICD10: Z33
Reference [1] Springer D, Zima T, Limanova Z. Reference intervals in evaluation of maternal thyroid function during the first trimester of pregnancy. European journal of endocrinology / European Federation of Endocrine Societies. 2009;160(5):791-7.
Tyreoid´ aln´ı stimulaˇ cn´ı hormon
[2] Lazarus JH, Kokandi A. Thyroid disease in relation to pregnancy: a decade of change. Clinical endocrinology. 2000;53(3):265-78.
Definice: Hormon vyluˇcovan´ y z pˇredn´ıho laloku hypof´ yzy a stimuluj´ıc´ı ˇst´ıtnou ˇzl´ azu.
[3] Glinoer D. The regulation of thyroid function in pregnancy: pathways of endocrine adaptation from physiology to pathology. Endocrine reviews. 1997;18(3):404-33.
Synonyma: Tyreotropin, Tyreotropn´ı hormon, hormon stimuluj´ıc´ı ˇst´ıtnou ˇzl´ azu, TSH
[4] Potlukova E, Potluka O, Jiskra J, Limanova Z, Telicka Z, Bartakova J, et al. Is age a risk factor for hypothyroidism in pregnancy? An analysis of 5223 pregnant women. The Journal of clinical endocrinology and metabolism. 2012;97(6):1945-52.
Zdroj: Merriam Webster: An Encyclopedia Britannica Company [online]. 2015 [cit. 2015-23-06]. Available from: http://www.merriam-webster.com/ SNOMED CT: 65428006 MeSH: D06.427.699.631.525.883; D12.644.548.691.525.883 ICD10: Y42.8
Protil´ atky proti tyreoid´ aln´ı peroxid´ aze Definice: Protil´atky nam´ıˇren´e proti tyreoid´ aln´ı peroxid´aze jsou enzym, norm´ alnˇe pˇr´ıtomn´ y ve ˇst´ıtn´e ˇzl´aze a hraj´ıc´ı d˚ uleˇzitou roli v produkci hormon˚ u ˇst´ıtn´e ˇzl´azy. Synonyma: anti-TPO, protil´ atky TPO, tyreoperoxid´azov´e protil´atky, protil´ atky proti peroxid´aze ˇst´ıtn´e ˇzl´azy
[5] Allan WC, Haddow JE, Palomaki GE, Williams JR, Mitchell ML, Hermos RJ, et al. Maternal thyroid deficiency and pregnancy complications: implications for population screening. Journal of medical screening. 2000;7(3):127-30. [6] Lazarus JH. Thyroid function in pregnancy. British medical bulletin. 2011;97:137-48. [7] Krassas GE, Poppe K, Glinoer D. Thyroid function and human reproductive health. Endocrine reviews. 2010;31(5):702-55. [8] Casey BM, Dashe JS, Wells CE, McIntire DD, Leveno KJ, Cunningham FG. Subclinical hyperthyroidism and pregnancy outcomes. Obstetrics and gynecology. 2006;107(2 Pt 1):337-41. [9] Benhadi N, Wiersinga WM, Reitsma JB, Vrijkotte TG, Bonsel GJ. Higher maternal TSH levels in pregnancy are associated with increased risk for miscarriage, fetal or neonatal death. European journal of endocrinology / European Federation of Endocrine Societies. 2009;160(6):985-91. [10] Stagnaro-Green A, Abalovich M, Alexander E, Azizi F, Mestman J, Negro R, et al. Guidelines of the American Thyroid Association for the diagnosis and management of thyroid disease during pregnancy and postpartum. Thyroid : official journal of the American Thyroid Association. 2011;21(10):1081-125.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Bart´akov´a J., Jiskra J. – Management n´asledn´ych vyˇsetˇren´ı ˇst´ıtn´e ˇz´azy v tˇehotenstv´ı
[11] Lazarus J, Brown RS, Daumerie C, Hubalewska-Dydejczyk A, Negro R, Vaidya B. 2014 European thyroid association guidelines for the management of subclinical hypothyroidism in pregnancy and in children. European thyroid journal. 2014;3(2):7694. [12] Burch HB, Burman KD, Cooper DS, Hennessey JV. A 2013 survey of clinical practice patterns in the management of primary hypothyroidism. The Journal of clinical endocrinology and metabolism. 2014;99(6):2077-85. [13] Leese B. Is there too much laboratory testing? Reprt 79. York,
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Great Britain: University of York. 1991:29pp. [14] Beck JR. Does feedback reduce inappropriate test ordering? Archives of pathology & laboratory medicine. 1993;117(1):334. [15] Bareford D, Hayling A. Inappropriate use of laboratory services: long term combined approach to modify request patterns. BMJ (Clinical research ed). 1990;301(6764):1305-7. [16] van Walraven C, Raymond M. Population-based study of repeat laboratory testing. Clinical chemistry. 2003;49(12):19972005.
5
6
Berger J., Beyr K. – Bezpeˇcnost osobn´ıch dat a Big Data v biomedic´ınˇe
Bezpeˇ cnost osobn´ıch dat a Big Data v biomedic´ınˇ e Jiˇr´ı Berger1 , Karel Beyr 1
´ ˇ a republika Ustav patologick´e fyziologie, Praha, Cesk´
Kontakt: Jiˇr´ı Berger ´ Ustav patologick´ e fyziologie Adresa: U Nemocnice 5, 128 53 Praha 2 E–mail:
[email protected]
C´ıle v´ yzkumu
hodnotu mohou m´ıt pro budouc´ı zpracov´an´ı r˚ uzn´ ych druh˚ u anal´ yz na z´akladˇe zdravotn´ı dokumentace cel´e poych informac´ı. Big Data skr´ yvaj´ı obrovsk´ y potenci´ al pro v´ yzkum pulace a souvisej´ıc´ıch biomedic´ınsk´ v oblasti biomedic´ıny v mnoha oblastech: • anal´ yza segmentace pacient˚ u, ceny a v´ ysledku l´eˇcby Jakmile dojde k nastaven´ı takov´eho projektu, aˇc umoˇzn´ı zjistit zdravotnˇe a cenovˇe nejefektivnˇejˇs´ı po- autoˇri tohoto ˇcl´anku pˇripouˇst´ı, ˇze vˇetˇs´ı neˇz technick´ y to stup l´eˇcen´ı pro konkr´etn´ıho pacienta; bude organizaˇcnˇe etick´ y probl´em, bude nejefektivnˇejˇs´ım zp˚ usobem pr´ace s takov´ ym mnoˇzstv´ım dat zpˇr´ıstupnˇen´ı • proaktivn´ı identifikace pacient˚ u, u nichˇz by se vyanonymizovan´ ych informac´ı odborn´e veˇrejnosti jako zdroj platila zdravotnick´ a prevence; v´ yzkumu. V zahraniˇc´ı je obvykl´e, ˇze pokud je projekt zcela ych • z anal´ yzy v´ yskytu chorob lze dˇelat epidemiologick´e nebo i jen ˇc´asteˇcnˇe financov´an a podporov´an z veˇrejn´ zdroj˚ u, b´ yvaj´ı nastaveny podm´ınky tak, aby byly inforz´avˇery a navrhovat preventivn´ı opatˇren´ı; mace dostupn´e k nekomerˇcn´ım aktivit´am s minim´aln´ımi • pomoc pˇri detekci a minimalizaci pokus˚ u o podvod omezen´ımi. I pˇresto, ˇze biomedic´ınsk´e informace, obve zdravotnictv´ı; zvl´aˇstˇe s ohledem na citlivost uchov´avan´ ych dat, nebude nikdy moˇzn´e poskytovat zcela bez omezen´ı, st´ale existuje • moˇznost spolupr´ace s farmaceutick´ ymi spoleˇcnostmi ˇsirok´a ˇsk´ala pouˇzit´ı, implementac´ı a aplikac´ı, pro kter´e by tak, ˇze pro nˇe bude moˇzno snadnˇeji identifikovat byl takov´ y pˇr´ıstup v´ yhodn´ y. skupinu relevantn´ıch pacient˚ u pro klinick´e testy (za pˇredpokladu pˇredchoz´ıho souhlasu pacient˚ u). V dneˇsn´ı dobˇe je patrn´ y trend digitalizace zdravotnick´ ych archiv˚ u a souvisej´ıc´ı dokumentace, nast´av´a tedy ˇcas na zapojen´ı technologi´ı oznaˇcovan´ ych jako Big Data v oblasti biomedic´ınsk´e informatiky. Tyto technologie nab´ızej´ı rychlejˇs´ı a efektivnˇejˇs´ı zpracov´an´ı a sd´ılen´ı obrovsk´eho mnoˇzstv´ı dat. Vzhledem k tomu, ˇze zdravotn´ı p´eˇce pracuje s velmi citliv´ ymi daty, je hlavn´ım z´ajmem ochrana dat pacient˚ u. V mnoha zem´ıch prob´ıh´a programov´e zav´ adˇen´ı elektronizace zdravotn´ı p´eˇce. Napˇr´ıklad v USA prob´ıh´ a Health Information ” Technology for Economic and Clinical Health Act“, (HITECH). C´ılem v´ yzkumu je n´ avrh a definice pravidel, kter´a zamez´ı zneuˇzit´ı a u ´nik˚ um citliv´ ych biomedic´ınsk´ ych dat. Souˇcasnˇe vˇsak v minim´ aln´ı m´ıˇre omez´ı efektivitu jejich zpracov´an´ı a kvalitu v´ ystupn´ıch dat.
To vˇsak, obzvl´aˇstˇe d´ıky pouˇzit´ı populaˇcn´ıch dat, s sebou pˇrin´aˇs´ı rizika moˇznosti nepˇr´ım´e identifikace konkr´etn´ıch informac´ı o pacientech. Sebemenˇs´ı n´aznak zneuˇzit´ı pˇrin´aˇs´ı etick´e probl´emy, a m˚ uˇze tak zastavit veˇsker´ y v´ yzkum.
Aby se pˇredeˇslo moˇzn´emu zneuˇzit´ı, je nutn´e nastavit velmi pˇr´ısn´a a efektivn´ı pravidla, umoˇzn ˇuj´ıc´ı maxim´aln´ı v´ ytˇeˇznost dat pˇri souˇcasn´em striktn´ım zachov´an´ı anonymity a ochrany osobn´ıch u ´daj˚ u. Souˇcasnˇe je nutn´e regulovat zp˚ usoby vytˇeˇzov´an´ı dat tak, aby nemohlo doj´ıt ke zneuˇzit´ı nebo u ´niku citliv´ ych informac´ı jak´ ymkoliv, i jen teoreticky provediteln´ ym, zp˚ usobem. V zahraniˇc´ı jiˇz existuje odpov´ıdaj´ıc´ı legislativa, napˇr. aktu´aln´ı verze Health ” Insurance Portability and Accountability Act“ (HIPAA) v USA, kter´a specifikuje standardy transakc´ı se zdraSouˇ casn´ y stav pozn´ an´ı votn´ımi z´aznamy, obdobnˇe smˇernice EU Data Protection Directive 95/46/EC, kter´a definuje poˇzadavek souhlasu ˇ ım vˇetˇs´ı mnoˇzstv´ı heterogenn´ıch biomedic´ınsk´ ´daj˚ u a pˇrenositelnost dat. C´ ych in- pacienta se zpracov´an´ım jeho u y pˇr´ıstup k ochranˇe citliv´ ych formac´ı se podaˇr´ı sdruˇzit pod technologie Big Data tak, EU vˇsak st´ale nem´a jednotn´ ´daj˚ u [1]. aby obsahovaly co nejkompletnˇejˇs´ı z´ aznamy, t´ım vyˇsˇs´ı u S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Berger J., Beyr K. – Bezpeˇcnost osobn´ıch dat a Big Data v biomedic´ınˇe
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı Vyuˇz´ıv´an´ı Big Data v biomedic´ınˇe a zdravotnictv´ı bude m´ıt vˇzdy sv´ a specifika. M´ıra anonymizace pouˇzit´ ych dat bude nepˇr´ımo u ´mˇern´ a kvalitˇe v´ ystup˚ u [2]. Znamen´a to, ˇze jedn´ım z kl´ıˇcov´ ych prvk˚ u u ´spˇeˇsn´eho vyuˇzit´ı Big Data v biomedic´ınˇe a zdravotnictv´ı bude nastaven´ı hranice mezi anonymizac´ı a potenci´ alem vytˇeˇzen´ ych dat. Pro efektivn´ı vyuˇzit´ı by bylo potˇrebn´e prov´est zcela z´akladn´ı anonymizaci, kter´ a z dat odstran´ı (nebo zaruˇcen´ ym zp˚ usobem znepˇr´ıstupn´ı) jen z´ akladn´ı osobn´ı informace, jako je jm´eno, pˇr´ıjmen´ı a rodn´e ˇc´ıslo, a nahrad´ı je anonymn´ım u ´dajem, kter´ y vˇsak pˇresto zajist´ı identifikaci subjektu napˇr´ıˇc daty. Takto upraven´ a data vˇsak budou st´ale zraniteln´a, a proto je nutn´e jednotliv´ ym zp˚ usob˚ um potenci´aln´ıch u ´tok˚ u efektivnˇe pˇredch´ azet. Napˇr´ıklad dotaz, kter´ y vrac´ı pˇredepsan´ a l´eˇciva a jejich d´avkov´an´ı konkr´etn´ımu pacientovi, obsahuje citliv´a data. Ze znalosti druhu l´eˇciv lze usuzovat na diagn´ ozu pacienta. Pokud budou osobn´ı u ´daje anonymizovan´e, lze usoudit, ˇze poskytnut´a data nebudou citliv´ a. Oproti tomu typick´ ymi dotazy, kter´e poskytuj´ı data, jenˇz nejsou citliv´ a, jsou napˇr´ıklad: Dotaz, kter´ y z´ısk´a poˇcet pacient˚ u dan´eho l´ekaˇre, nemus´ı nutnˇe obsahovat citliv´a data. Stejnˇe tak dotaz na seznam pˇredepsan´ ych l´eˇciv v dan´em regionu nevrac´ı citliv´e v´ ysledky. Podobnˇe v´ yskyt specifick´e diagn´ ozy napˇr´ıˇc populac´ı neposkytuje citliv´e u ´daje.
Bezpeˇ cnost datab´ azov´ ych dat jako celku
Nepˇr´ım´ e z´ısk´ an´ı konkr´ etn´ıch dat z ˇ c´ asteˇ cnˇ e anonymizovan´ eho souboru dat V pˇr´ıpadˇe, ˇze se v datab´azi nach´azej´ı kompletn´ı neanonymizovan´a data nebo ˇcastˇeji data, kter´a proˇsla pouze z´akladn´ı anonymizac´ı, v r´amci n´ıˇz byla nahrazena z´akladn´ı data pacient˚ u (jm´eno, pˇr´ıjmen´ı, rodn´e ˇc´ıslo), ale zbytek datov´eho souboru je kompletn´ı, je nutn´e detailnˇe ˇreˇsit principy omezen´ı pˇr´ıstupu. V takov´em pˇr´ıpadˇe lze pˇri c´ılen´em u ´toku kombinac´ı dotaz˚ u z´ıskat velmi konkr´etn´ı data, nebo pˇrinejmenˇs´ım data kter´a lze s vysokou pravdˇepodobnost´ı interpretovat zcela konkr´etnˇe. Z v´ yˇse uveden´ ych d˚ uvod˚ u je nutn´e zav´est takov´e koncepˇcn´ı ˇreˇsen´ı, jeˇz spoˇc´ıv´a v omezen´ı dotaz˚ u, jejichˇz kombinace m˚ uˇze odhalit citliv´e u ´daje, pˇr´ıpadnˇe takov´e kombinace umoˇznit jen osob´am s vyˇsˇs´ım opr´avnˇen´ım a souˇcasnˇe zajistit zpˇetnou kontrolu a anal´ yzu rizikovosti pouˇz´ıvan´ ych dotaz˚ u a jejich v´ ysledk˚ u. Dalˇs´ım rizikov´ ym dotazem je takov´ y, kter´ y ve sv´em v´ ysledku pˇred´a v´ yznamnˇe mal´ y poˇcet v´ ystupn´ıch entit. Dotazy mohou obsahovat kombinaci nˇekolika faktor˚ u. Pokud je ale dan´ y dotaz pˇrekombinov´an“, m˚ uˇze doj´ıt ” v extr´emn´ım pˇr´ıpadˇe k situaci, ˇze jeho v´ ysledkem bude pouze jeden pacient, u kter´eho i bez znalosti jm´ena m˚ uˇzeme odvodit, o koho se jedn´a. Napˇr´ıklad pokud zn´ame i pouhou ˇc´ast chorobopisu dan´eho ˇclovˇeka, lze pomoc´ı souvisej´ıc´ıch informac´ı (vˇek, pohlav´ı, bydliˇstˇe) nepˇr´ımo z´ıskat jeho citliv´e informace. Takov´e riziko se d´a do urˇcit´e m´ıry eliminovat nasazen´ım heuristick´ ych pravidel a jejich postupn´ ym zpˇresˇ nov´an´ım, a t´ım zablokovat odpovˇedi, kter´e by mohly obsahovat rizikovou mnoˇzinu informac´ı.
Nepˇr´ım´ e z´ısk´ an´ı konkr´ etn´ıch dat z plnˇ e anonymizovan´ eho souboru dat
Jednou z metod zv´ yˇsen´ı bezpeˇcnosti dat v biomedic´ınˇe V pˇr´ıpadˇe volby varianty anonymizace datov´eho soua l´ekaˇrstv´ı je ˇsifrov´ an´ı podkladov´ ych dat, kter´e pˇrid´av´a boru lze za stanoven´ ych podm´ınek umoˇznit pln´ y pˇr´ıstup dalˇs´ı bezpeˇcnostn´ı vrstvu nav´ıc, a pr´ avˇe n´ avrh t´eto archi- do datab´aze. tektury umoˇzn ˇuje v´ yznamnˇe sn´ıˇzit moˇznosti zneuˇzit´ı dat Pˇred uloˇzen´ım dat do datab´aze je moˇzno (pˇr´ıpadnˇe [3]. nutno, podle dan´e legislativy) data anonymizovat (odstran generalizovat. Tomuto Existuj´ı specializovan´e pokroˇcil´e algoritmy[4] nit jm´eno, rodn´e ˇc´ıslo) a z´aroveˇ zp˚ u sobu ˇ r ´ ık´ a me pln´ a anonymizace. Generalizace spoˇc´ıv´a umoˇzn ˇuj´ıc´ı ˇsifrov´ an´ı medic´ınsk´ ych z´ aznam˚ u tak, ˇze ve znesnadnˇ e n´ ı identifikace osoby pomoc´ı tzv. kvaziumoˇzn´ı jejich rozˇsifrov´ an´ı pouze osob´ am s relevantn´ım ” identifik´ a tor˚ u “ – to jsou napˇ r ´ ıklad datum narozen´ı, adresa opr´avnˇen´ım. Algoritmy maj´ı nˇekolik v´ yhod oproti klau unikl na veˇrejnost sick´ ym ˇsifrovac´ım zp˚ usob˚ um (symetrick´e i asymetrick´e a pohlav´ı. Pomoc´ı kvazi-identifik´ator˚ napˇ r . zdravotn´ ı stav guvern´ e ra st´ a tu Massachusetts pot´e, ˇsifry) – oproti RSA konceptu jsou rychlejˇs´ı, levnˇejˇs´ı a flexico byly zveˇ r ejnˇ e ny jeho kvazi-identifik´ atory z volebn´ıho bilnˇejˇs´ı, oproti symetrick´ ym ˇsifr´ am poskytuj´ı bezpeˇcnost z´ a znamu, kter´ e odpov´ ıdaly jeho anonymizovan´ emu zdrav pˇr´ıpadˇe vyzrazen´ı sd´ılen´eho hesla. V [5] popisuj´ı votn´ ımu z´ a znamu. autoˇri moˇznosti vyhled´ av´ an´ı nad takto ˇsifrovan´ ymi meV [6] popisuj´ı autoˇri generalizaci dat zaloˇzenou na tom dic´ınsk´ ymi daty. Tento algoritmus je ide´ aln´ı pro koncept principu, ˇze nesm´ı existovat mnoˇziny pacient˚ u maj´ıc´ıch Big Data. stejn´e kvazi-identifik´atory obsahuj´ıc´ı m´enˇe neˇz K prvk˚ u. ˇ Sifrov´ an´ı pˇrinese zpomalen´ı vyhled´ av´ an´ı, proto jej Tento pˇr´ıstup znesnadˇ nuje identifikaci osob, ale byly nelze doporuˇcit pro pouˇzit´ı pro cel´e u ´loˇziˇstˇe, n´ ybrˇz pouze zjiˇstˇeny pˇr´ıpady, kdy i pˇres K-anonymizaci bylo moˇzn´a pro z´akladn´ı data pacient˚ u, u kter´ ych je jasnˇe dan´a struk- identifikace. V [7] autoˇri vych´az´ı z K-anonymizace a natura. vrhuj´ı zlepˇsen´ı, L-anonymizaci. Ta spoˇc´ıv´a v tom, ˇze S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
7
8
Berger J., Beyr K. – Bezpeˇcnost osobn´ıch dat a Big Data v biomedic´ınˇe
vyˇzaduje r˚ uznorodost citliv´ ych u ´daj˚ u v mnoˇzinˇe osob se • Z´avislost v´ yskytu chorob na druhu zamˇestn´an´ı pastejn´ ymi kvazi-identifik´ atory. cienta. Dalˇs´ı moˇznost´ı generalizace je znepˇresnˇen´ı kvaziPro tuto potˇrebu lze vybudovat nˇekolikavrstvou archiidentifik´ator˚ u. Stejn´a data mohou b´ yt v datab´ azi uloˇzena tekturu, kter´a bude nadstavbou z´akladn´ıch Big Data technˇekolikr´at, pokaˇzd´e s r˚ uzn´ ym stupnˇem pˇresnosti, s t´ım, ˇze nologi´ı v konkr´etn´ı biomedic´ınsk´e implementaci a bude ˇc´ım vˇetˇs´ı m´a ˇcten´aˇr opr´ avnˇen´ı, t´ım k pˇresnˇejˇs´ım dat˚ um rozdˇelena nejm´enˇe do tˇechto vrstev: se m˚ uˇze dostat. Napˇr. m´ısto data narozen´ı se uchov´av´a pouze rok nebo dokonce jen dek´ ada, m´ısto adresy pouze 1. Odborn´a veˇrejnost bude moci pouˇz´ıvat pouze n´azev obce ˇci kraje. pˇredpˇripraven´e ˇsablony, do kter´ ych bude moˇzn´e Jin´a moˇznost generalizace spoˇc´ıv´ a v tom, ˇze nˇekter´e vkl´adat vlastn´ı parametry, ale nebude moˇzn´e mˇenit kvazi-identifik´atory nebudou v datab´ azi v˚ ubec dostupn´e. podstatu dotaz˚ u. Dotaz p˚ ujde pokl´adat pouze jedV zahraniˇc´ı se pro popis chorob a symptom˚ u pouˇz´ıvaj´ı nou ˇsablonou, ˇsablony nelze kombinovat. Takto ICD k´ody (International Statistical Classification of Dibude zaruˇceno, ˇze nedojde k u ´niku citliv´ ych u ´daj˚ u. seases and Related Health Problems) spadaj´ıc´ı pod Tento pˇr´ıstup bude slouˇzit napˇr. postgradu´aln´ım spr´avu WHO. Tyto k´ ody maj´ı hierarchickou strukturu, student˚ um pro jejich z´akladn´ı v´ yzkum. a tud´ıˇz se pˇr´ımo nab´ızej´ı ke generalizaci. Autoˇri [8] uv´ad´ı 2. Specializovan´a pracoviˇstˇe budou schopna kombinopravdˇepodobnost identifikace pacienta podle ˇcetnosti ravat ˇsablony a v´ıce parametrizovat jednotliv´e dotazy. ritn´ıch ICD k´od˚ u. Doporuˇcuj´ı odstranit 5% aˇz 25% raSouˇcasnˇe vˇsak bude nad jejich ˇcinnost´ı bd´ıt sada ritn´ıch k´od˚ u (symptom˚ u) a nahradit je jejich generalizac´ı. heuristick´ ych pravidel, jeˇz bude reportovat nebo ve T´ım se v´ yznamnˇe sn´ıˇz´ı pravdˇepodobnost identifikace pavybran´ y ch pˇr´ıpadech i blokovat pouˇzit´ı kombinace cienta, za relativnˇe n´ızkou cenu ztr´ aty pˇresnosti dat. dotaz˚ u, kter´e mohou pˇrin´aˇset riziko u ´niku konkretizovateln´ ych dat.
Netrivi´ alnost dotazov´ an´ı
Velk´ y probl´em t´ ykaj´ıc´ı se v´ yzkumu nad Big Data v biomedic´ınsk´e informatice spoˇc´ıv´ a ve vlastn´ı tvorbˇe dotaz˚ u. Nelze pˇredpokl´adat, ˇze by v´ yznamn´ a vˇetˇsina vˇedeck´ ych pracovn´ık˚ u zab´ yvaj´ıc´ı se biomedic´ınsk´ ym oborem byla schopna a ochotna programovat vlastn´ı Map/Reduce paraleln´ı programy pro ˇreˇsen´ı dotaz˚ u nad Big Data medic´ınskou datab´az´ı. Sp´ıˇse bude pravdˇepodobn´ ym sc´en´aˇrem nastaven´ı takov´e spolupr´ ace, kdy informatici, analytici nebo program´ atoˇri vytvoˇr´ı n´ astroje, kter´e bude moˇzn´e parametrizovat, spouˇstˇet apod. Obˇe v´ yˇse popsan´e oblasti (bezpeˇcnost a netrivialita dotazov´an´ı) je moˇzn´e vyˇreˇsit pomoc´ı dotazovac´ıho n´astroje. N´astroj by obsahoval ˇsablony“ dotaz˚ u, program˚ u ” nebo algoritm˚ u, kter´e by se daly pˇri pouˇzit´ı uˇzivatelsk´eho rozhran´ı parametrizovat a takto pouˇz´ıt k prohled´av´an´ı datab´aze odbornou veˇrejnost´ı, aniˇz by musela proch´azet n´aroˇcn´ ym procesem ˇskolen´ı a v´ yuky pouˇz´ıv´ an´ı Big Data. Pouˇzit´ı ˇsablon by jako prim´ arn´ı c´ıl mˇelo v´ yraznˇe usnadnit pouˇz´ıv´an´ı ˇsirˇs´ı vˇedeck´e komunitˇe a zv´ yˇsit dostupnost takto zamˇeˇren´eho v´ yzkumu a jeho v´ ystup˚ u pro ˇsirok´e spektrum aplikac´ı. Z´ aroveˇ n by se t´ımto zp˚ usobem elegantnˇe vyˇreˇsila ot´azka bezpeˇcnosti. Pˇr´ıklad ˇsablon: • Preskripce konkr´etn´ı u ´ˇcinn´e l´ atky dle l´ekaˇrsk´e specializace; • Anal´ yza ˇcetnosti podle m´ısta trval´eho pobytu pacienta;
3. Analytick´ y t´ ym bude pˇripravovat ˇsablony vˇcetnˇe heuristick´ ych pravidel, kter´a budou pouˇz´ıv´an´ı takov´ ych ˇsablon hl´ıdat. Souˇcasnˇe bude pod standardn´ımi bezpeˇcnostn´ımi kontrolami (auditn´ı log, anal´ yza ˇcetnosti v´ ysledk˚ u apod.) schopen Big Data s´am vyuˇz´ıvat v pˇr´ıpadˇe, ˇze bude existovat vysok´e riziko pr´ace s citliv´ ymi daty. V takov´em t´ ymu budou lid´e s odpov´ıdaj´ıc´ım provˇeˇren´ım, na kter´e budou aplikovan´e i procesn´ı postupy zajiˇst’uj´ıc´ı bezpeˇcnost vyuˇz´ıv´an´ı dat. Tento t´ ym pak m˚ uˇze jako souˇc´ast sv´e n´aplnˇe pr´ace zpracov´avat komplexn´ı u ´lohy a dotazy dle poˇzadavk˚ u jednotliv´ ych pracoviˇst’ v pˇr´ıpadech, kdy nebude efektivn´ı pouˇz´ıt schematick´e ˇsablony a souˇcasnˇe bude existovat riziko u ´niku citliv´ ych dat. Z´ıskan´a data budou pˇred pˇred´an´ım ˇz´adaj´ıc´ımu pracoviˇsti zkontrolov´ana a v pˇr´ıpadˇe potˇreby bude soubor v´ ysledk˚ u dodateˇcnˇe anonymizov´an. ´ y specializovan´ 4. Uzk´ y auditn´ı t´ ym bude nastavovat pokroˇcil´a heuristick´a pravidla nad cel´ ym syst´emem, schvalovat ˇsablony pˇred zveˇrejnˇen´ım a definovat bezpeˇcnostn´ı pravidla. 5. Posledn´ım prvkem m˚ uˇze v budoucnosti b´ yt syst´em pracuj´ıc´ı s prvky umˇel´e inteligence, kter´ y bude na z´akladˇe souboru informac´ı s vyuˇzit´ım pokroˇcil´ ych algoritm˚ u rozpozn´av´an´ı vzorc˚ u, neuronov´ ych s´ıt´ı a uˇc´ıc´ıho se procesu automatizovanˇe vyhled´avat a vyhodnocovat nepodchycen´e principy a moˇznosti zneuˇzit´ı dat v re´aln´em ˇcase.
Diskuze
• Demografick´a skladba pacient˚ u; • Sez´onn´ı objemy v´ ykon˚ u medic´ınsk´ ych zaˇr´ızen´ı;
Pokraˇcov´an´ı v´ yzkumu bude zamˇeˇreno na nˇekolik kl´ıˇcov´ ych oblast´ı. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Berger J., Beyr K. – Bezpeˇcnost osobn´ıch dat a Big Data v biomedic´ınˇe
Prvn´ı z nich je zamˇeˇren´ı na obvykl´ a bezpeˇcnostn´ı pravidla a souvisej´ıc´ı technologie zabezpeˇcen´ı dat, jejich vyuˇzit´ı v r´amci biomedic´ınsk´ ych dat a definice specifik a odchylek pˇr´ıstupu k biomedic´ınsk´ ym dat˚ um a zdravotn´ım z´aznam˚ um oproti standardnˇe vyuˇz´ıvan´ ym metod´am ochrany dat. Do t´eto oblasti patˇr´ı hlavnˇe anal´ yza princip˚ u ˇsifrovac´ıch algoritm˚ u dle udˇelen´ ych opr´avnˇen´ı, jejich pˇr´ınosy a nev´ yhody a v neposledn´ı ˇradˇe i vztah r˚ uzn´ ych princip˚ u ˇsifrov´ an´ı k moˇznostem a efektivitˇe anal´ yzy dat. Dalˇs´ı oblast´ı bude v´ yzkum m´ıry anonymizace s ohledem na v´ ytˇeˇznost a efektivitu jejich zpracov´an´ı a definice princip˚ u a vlivu anonymizace na biomedic´ınsk´a data s ohledem na teoretick´e moˇznosti nepˇr´ım´eho z´ısk´an´ı konkr´etn´ıch dat za pomoc´ı kombinace pokl´ adan´ ych dotaz˚ u. Tato ˇsirok´ a problematika bude rozpracov´ana jak v teoretick´e rovinˇe, tak dojde ke zobecnˇen´ı konkr´etn´ıch pˇr´ıpad˚ u z´ıskan´ ych kombinatorick´ ymi metodami z reprezentativn´ıho vzorku anonymizovan´ ych dat pˇri jejich porovn´an´ı se skuteˇcn´ ymi daty a s vyhodnocen´ım m´ıry shody. Nejrozs´ahlejˇs´ı oblast´ı v´ yzkumu je definice rozhran´ı mezi program´ atorsk´ ymi metodami a moˇznosti vyuˇzit´ı ˇsablon dotaz˚ u, kde hlavn´ım c´ılem bude nalezen´ı hranice, kter´ a zajist´ı vysokou m´ıru kontroly pˇri souˇcasn´em zpˇr´ıstupnˇen´ı dotazovac´ıch metod ˇsirok´e odborn´e veˇrejnosti v takov´e podobˇe, kdy jednotliv´ı koncov´ı uˇzivatel´e budou schopni bez programov´ an´ı a v´ yznamnˇejˇs´ı m´ıry zaˇskolen´ı prov´ adˇet vlastn´ı v´ yzkum nad daty. C´ılov´ ym stavem je nalezen´ı rozhran´ı, kter´e pro odbornou veˇrejnost bude svou pouˇzitelnost´ı pˇrirovnateln´e k pokroˇcilejˇs´ımu pouˇz´ıv´ an´ı kancel´ aˇrsk´ ych softwarov´ ych bal´ık˚ u, jak´ ymi jsou napˇr. MS Excel nebo MS Access, pˇr´ıpadnˇe jejich alternativy. V r´ amci t´eto oblasti bude analyzov´ano a upˇresnˇeno rozdˇelen´ı pr´ av a zodpovˇednost´ı do navrˇzen´ ych ˇctyˇr rol´ı, jejich podrobn´ y popis a procesn´ı zmapov´an´ı vztah˚ u tˇechto rol´ı k zajiˇstˇen´ı bezpeˇcnosti dat. V r´amci jednotliv´ ych rol´ı budou analyzov´ ana potenci´aln´ı rizika a hrozby vˇcetnˇe princip˚ u, jak jim pˇredch´ azet. V z´akladn´ı podobˇe bude rozpracov´ an i princip vyuˇzit´ı heuristick´ ych anal´ yz a prvk˚ u umˇel´e inteligence jako n´ astroje pro zv´ yˇsen´ı ochrany biomedic´ınsk´ ych dat a zdravotnick´ ych z´aznam˚ u.
Podˇ ekov´ an´ı Tato pr´ace byla podpoˇrena projektem SVV-2015260158 Univerzity Karlovy v Praze.
SNOMED CT: nenalezeno MeSH: nenalezeno ICD10: nenalezeno
Big Data Definice: Big Data je sada prostˇredk˚ u a technologi´ı, kter´e vyˇzaduj´ı nov´e formy integrace k odkryt´ı skryt´ ych vazeb z datab´az´ı, kter´e jsou rozmanit´e, komplexn´ı a rozs´ahl´e. Synonyma: Cloudov´e u ´loˇziˇstˇe, nˇekolikaserverov´a datab´aze Zdroj: [6] SNOMED CT: nenalezeno MeSH: nenalezeno ICD10: nenalezeno
Anonymizace Definice: Proces, kter´ ym se zaˇsifruj´ı nebo odstran´ı osobn´ı u ´daje z mnoˇziny dat tak, ˇze lid´e, kter´e dan´a data popisuj´ı, z˚ ustanou anonymn´ı. Synonyma: Oˇciˇstˇen´ı dat od osobn´ıch u ´daj˚ u Zdroj: [6] SNOMED CT: nenalezeno MeSH: nenalezeno ICD10: nenalezeno
Reference [1] Boussi Rahmouni H, Solomonides T, Casassa Mont M, Shiu S. Modelling and Enforcing Privacy for Medical Data Disclosure across Europe. In Adlassnig KP, editor. Medical Informatics in a United and Healthy Europe – Proceedings of. Sarajevo: IOS Press ; 2009. p. 695-699. [2] Duncan et al. Disclosure Risk vs. Data Utility: The R-U Confidentiality map: Los Alamos National Library; 2001. [3] Amazon Web Services. Creating Healthcare Data Applications to Promote HIPAA and HITECH Compliance. 2012.
Kl´ıˇ cov´ a slova
[4] Alshehri , Radziszowski , Raj K. Designing a Secure CloudBased EHR System using Ciphertext-Policy Attribute-Based Encryption.
Osobn´ı u ´daje
[5] Narayan S, Gagn´ e M, Reihaneh SN. Privacy preserving EHR system using attribute-based infrastructure.
Definice: Veˇsker´e informace o identifikovan´e nebo identifikovateln´e osobˇe.
[6] Sweeney L. k-anonymity: a model for protecting privacy. International Journal on Uncertainty. 2002; 10(5): p. 557-570.
Synonyma: Data identifikuj´ıc´ı osobu
[7] Machanavajjhala A, Kifer D, Gehrke J, Venkitasubramaniam M. L-diversity: Privacy beyond k-anonymity. ACM Transactions on Knowledge Discovery from Data. 2007 March; 1(1).
Zdroj: ˇcl. 2 p´ısm. a) smˇernice ˇc. 95/46/ES o ochranˇe fyzick´ ych osob v souvislosti se zpracov´ an´ım osobn´ıch u ´daj˚ u a o voln´em pohybu tˇechto u ´daj˚ u
[8] Vinterbo S, L OM, S D. Hiding information by cell suppression. In Proc AMIA Symp; 2001. p. 726–730.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
9
10
Bielnikov´a H. a kol. – Role jednonukleotidov´ych polymorfism˚ u v patogenezi vysokostupˇ nov´ych gliom˚ u
Role jednonukleotidov´ ych polymorfism˚ u v patogenezi vysokostupˇ nov´ ych gliom˚ u Hana Bielnikov´ a1,2 , Petr Buzrla1 , Ondˇrej Bielnik3 , Radoslava Tomanov´ a1 , Irena Urbanovsk´ a4 , Lucie Hruˇskov´ a2 , Jana Dvoˇr´ aˇ ckov´ a1 , Ivan Mazura2
3
1
´ ˇ a republika Ustav patologie, Fakultn´ı nemocnice Ostrava, Ostrava, Cesk´
2
ˇ a republika 1. l´ekaˇrsk´ a fakulta, Karlova univerzita v Praze, Praha, Cesk´
ˇ a republika Neurochirurgick´ a klinika, Fakultn´ı nemocnice Ostrava, Ostrava, Cesk´ 4
ˇ a republika CGB laboratoˇr Ostrava a.s., Ostrava, Cesk´
Kontakt: Hana Bielnikov´ a ´ Ustav patologie, Fakultn´ı nemocnice Ostrava Adresa: 17. listopadu 1790, 70852, Ostrava-Poruba E–mail:
[email protected]
C´ıle v´ yzkumu Jednonukleotidov´e polymorfismy (SNP) jsou odchylky jednotliv´ ych nukleotid˚ u v sekvenci deoxyribonukleov´e kyseliny (DNA) mezi dvˇema jedinci. Jde o jedny z nejˇcastˇeji se vyskytuj´ıc´ıch genetick´ ych zmˇen v k´ oduj´ıc´ıch ´ cinky i nek´oduj´ıc´ıch oblastech lidsk´eho genomu [1]. Uˇ tˇechto zmˇen se projevuj´ı v z´ avislosti na lokalizaci polymorfism˚ u v DNA. Pokud se SNP nach´ az´ı v nek´ oduj´ıc´ıch oblastech, jejich vliv na organismus je obvykle zanedbateln´ y a vˇetˇsinou nevede k fenotypov´ ym projev˚ um. Pˇri v´ yskytu polymorfism˚ u v k´oduj´ıc´ıch oblastech je jejich pˇr´ıpadn´ y dopad na fenotyp d´an ˇradou podm´ınek. C´ılem m´e pr´ace je izolace DNA z n´ adorov´e tk´anˇe pacient˚ u s diagn´ozou astrocytomu nebo glioblastomu a n´asledn´a anal´ yza vybran´ ych gen˚ u pomoc´ı molekul´arnˇegenetick´ ych metod. Z´ıskan´ a fenotypov´ a, klinick´ a a genetick´a data budou posuzov´ ana z hlediska jejich moˇzn´e korelace.
Souˇ casn´ y stav pozn´ an´ı Gliomy mozku jsou jedny z nejˇcastˇejˇs´ıch prim´ arn´ıch n´ador˚ u mozkov´e tk´anˇe dospˇel´ ych. Progn´ oza tˇechto tumor˚ u, v z´avislosti na jejich stupni, je velmi ˇspatn´a. U malign´ıch typ˚ u jako jsou anaplastick´ y astrocytom (AA, stupeˇ n III.) a glioblastom (GBM, stupeˇ n IV.) se doba pˇreˇz´ıv´an´ı pohybuje v rozmez´ı 3 aˇz 5 let u AA a 15 aˇz 16 mˇes´ıc˚ u u GBM [2]. Nebezpeˇc´ım tˇechto tumor˚ u je jejich infiltrativn´ı r˚ ust do mozkov´e tk´ anˇe znesnadˇ nuj´ıc´ı jejich chirurgick´e odstranˇen´ı, vysok´ a proliferaˇcn´ı aktivita, atypie jader, hojn´a angiogeneze a pˇr´ıtomnost nekr´ oz [3]. Zlat´ ym standardem pro diagnostiku gliom˚ u je histologick´e vyˇsetˇren´ı, kter´e je vˇsak komplikov´ ano znaˇcnou he-
terogenitou tumor˚ u a pˇrekr´ yvaj´ıc´ımi se morfologick´ ymi znaky, zvl´aˇstˇe u vyˇsˇs´ıch stupˇ n˚ u. Z tohoto d˚ uvodu jsou v posledn´ıch letech st´ale v´ıce v diagnostice a l´eˇcbˇe gliom˚ u vyuˇz´ıv´any poznatky z oblasti genetiky. Pomoc´ı cytogenetick´ ych a molekul´arnˇe-genetick´ ych metod byla objevena ˇrada alterac´ı chromozom˚ u a gen˚ u, kter´e maj´ı v´ yznamnou roli v upˇresnˇen´ı typizace tumor˚ u nebo jsou d˚ uleˇzit´ ymi prediktivn´ımi a prognostick´ ymi faktory [4]. Mnoˇzstv´ı genetick´ ych zmˇen, kter´e se u gliom˚ u vyskytuj´ı, je znaˇcnˇe ˇsirok´e. Jedn´a se o zmˇeny jak na u ´rovni cel´ ych chromozom˚ u a jejich ˇc´ast´ı (aneuploidie, delece, duplikace, translokace aj.), tak na u ´rovni vl´akna DNA (bodov´e mutace, delece, inzerce, amplifikace, substituce, epigenetick´e zmˇeny – metylace, acetylace aj.). Vˇetˇsina tˇechto alterac´ı se t´ yk´a oblast´ı, ve kter´ ych se vyskytuj´ı d˚ uleˇzit´e tumor-supresorov´e nebo protoonkogenn´ı geny [5, 6]. V posledn´ıch letech se ˇrada v´ yzkum˚ u soustˇred’uje na hled´an´ı souvislost´ı mezi genetick´ ymi polymorfismy a vznikem onkologick´ ych onemocnˇen´ı. Jedn´ım z diskutovan´ ych polymorfism˚ u, tak´e u gliom˚ u, je SNP309 (rs2279744) v promotorov´e oblasti genu MDM2. V pˇr´ıpadˇe polymorfismu SNP309 doch´az´ı k zv´ yˇsen´e schopnosti transkripˇcn´ıho aktiv´atoru Sp1 nav´azat se na DNA a t´ım zv´ yˇsit expresi genu MDM2 [7]. Produktem genu MDM2 je protein reguluj´ıc´ı dr´ahu p53, jednoho z nejd˚ uleˇzitˇejˇs´ıch tumor-supresorov´ ych faktor˚ u. Protein MDM2 se jako regul´ator pˇr´ımo v´aˇze na p53 a negativnˇe ovlivˇ nuje jeho stabilitu a aktivitu. Bylo prok´az´ano, ˇze i nepatrn´a zmˇena hladiny MDM2 v´ yraznˇe ovlivˇ nuje supresorov´e funkce p53 a u nˇekter´ ych typ˚ u n´ador˚ u je zv´ yˇsen´a exprese tohoto genu spojov´ana s rychlejˇs´ı progres´ı onemocnˇen´ı a slabˇs´ı odezvou na l´eˇcbu [8]. Negativnˇe mohou p˚ usobit tak´e SNP v k´oduj´ıc´ıch oblastech DNA, kdy v nˇekter´ ych situac´ıch z´amˇenou b´aze S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Bielnikov´a H. a kol. – Role jednonukleotidov´ych polymorfism˚ u v patogenezi vysokostupˇ nov´ych gliom˚ u
v kodonu doch´az´ı k v´ ymˇenˇe jedn´e aminokyseliny za jinou. Zdroj: http://lekarske.slovniky.cz/pojem/gliom Negativn´ım d˚ usledkem tˇechto z´ amˇen m˚ uˇze b´ yt vznik defektn´ıho nebo nefunkˇcn´ıho proteinu, pˇr´ıpadnˇe vznik stop SNOMED CT: 67271001 kodonu, jehoˇz n´asledkem je vytvoˇren´ı zkr´ acen´eho proteiMeSH: D005910 nov´eho produktu, ˇcasto nefunkˇcn´ıho.
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı
ICD10: nenalezeno
Jednonukleotidov´ y polymorfismus
Vˇetˇsina polymorfism˚ u pravdˇepodobnˇe pˇr´ım´ y vliv na zdrav´ı ˇclovˇeka nem´ a. Podstatn´ a je lokalizace tˇechto zmˇen v r´amci sekvence nukleov´e kyseliny. V nˇekter´ ych pˇr´ıpadech jejich pˇr´ıtomnost m˚ uˇze pozmˇenit u ´ˇcinnost l´ek˚ u, ovlivnit reakci organismu na vystaven´ı ˇskodlivin´am nebo b´ yt asociov´ana se vznikem a rozvojem nˇekter´ ych onemocnˇen´ı. Sledov´an´ı souvislost´ı mezi polymorfismy a gliomy n´am m˚ uˇze pˇrin´est nov´e poznatky o vzniku a rozvoji tˇechto n´ador˚ u, kdy nˇekter´e z vyˇsetˇrovan´ ych SNP by mohly slouˇzit jako potencion´ aln´ı prognostick´e a diagnostick´e ukazatele.
Definice: Polymorfismus sekvence DNA dan´ y variabilitou pouze v jedn´e b´azi – z´amˇena v jednom nukleotidu DNA, tato z´amˇena mus´ı b´ yt v populaci rozˇs´ıˇrena.
Podˇ ekov´ an´ı
ICD10: nenalezeno
Synonyma: Polymorfismus, SNP Zdroj: http://www.genomia.cz/cz/slovnik-pojmu/ SNOMED CT: nenalezeno MeSH: D020641
Studie byla podpoˇrena projektem SVV 260158 Univer- Genetick´ a mutace zity Karlovy v Praze.
Kl´ıˇ cov´ a slova Deoxyribonukleov´ a kyselina
Definice: Zmˇena genetick´e dˇediˇcn´e informace na u ´rovni DNA t´ ykaj´ıc´ı se bud’ gen˚ u, nebo cel´ ych chromozom˚ u. Podle m´ısta, kter´e je zasaˇzeno, m˚ uˇze ˇci nemus´ı ovlivˇ novat funkci buˇ nky a organismu. Vznik´a samovolnˇe, nebo je zp˚ usobena zevn´ımi faktory – mutageny chemick´e, fyzik´aln´ı nebo biologick´e vlivy.
Definice: Dvouvl´ aknov´ y polynukleotid tvoˇren´ y dvˇema samostatn´ ymi ˇretˇezci deoxyribonukleotidov´ ych jedSynonyma: Mutace notek. Slouˇz´ı jako nosiˇc genetick´e informace. Synonyma: DNA, DNK
Zdroj: http://lekarske.slovniky.cz/pojem/mutace
SNOMED CT: 55446002 ´ Zdroj: Alberts B, et al. Z´ aklady bunˇeˇcn´e biologie: Uvod do molekul´ arn´ı biologie buˇ nky. Espero Publishing MeSH: D009154 1998. SNOMED CT: 24851008
ICD10: nenalezeno
MeSH: D004247
Fenotyp
ICD10: nenalezeno
Gliom
Definice: Pozorovateln´ y vzhled ˇci vlastnost jedince, kter´ y je v´ ysledkem jeho dˇediˇcn´ ych vloh genotypu a p˚ usoben´ı prostˇred´ı.
Definice: N´ador CNS vych´ azej´ıc´ı z podp˚ urn´e nervov´e tk´anˇe neuroglie, glie. Patˇr´ı sem astrocytom, zhoubn´ y glioblastom aj. Gliomy rostou v r˚ uzn´ ych ˇc´ astech mozku. Zp˚ usobuj´ı epilepsii, psychick´e zmˇeny, r˚ ust nitrolebn´ıho tlaku s bolest´ı hlavy a poruchou vidˇen´ı, loˇziskov´e pˇr´ıznaky atd. Existuj´ı ve zhoubn´e i nezhoubn´e variantˇe.
Synonyma:
Synonyma: N´ador gli´ aln´ıch bunˇek
ICD10: nenalezeno
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Zdroj: http://lekarske.slovniky.cz/pojem/fenotyp SNOMED CT: 363778006 MeSH: D010641
11
12
Bielnikov´a H. a kol. – Role jednonukleotidov´ych polymorfism˚ u v patogenezi vysokostupˇ nov´ych gliom˚ u
Reference
[5] Hersh DS, Mehta RI, et al.The Molecular Pathology of Primary Brain Tumors. Path C Rev. 2013 Sept;18(5):210-220.
[1] Clark DP., Molecular biology: Understanding the Genetic Revolution. Elsevier Academic Press, 2005.
[6] Duncan ChG, Yan H. Genomic alterations and the pathogenesis of glioblastoma. Cell Cycle. 2011 Apr;10(8):1174-1175.
[2] Karajannis MA, Zagzag D., editors. Molecular pathology of nervous system tumors. New York, Springer, 2015.
[7] Wan Y, Wu W, Yin Z, Guan P, Zhou B. MDM2 SNP309, genegene interaction, and tumor susceptibility: an updated metaanalysis. BMC Can.2011;11(208):1-9.
[3] Iacob G, Dinca EB. Current data and strategy in glioblastoma multiforme. J Med Lif. 2009 Agu;2(4):386-393. [4] Nikiforova MN, Hamilton RL. Molecular diagnostics of gliomas. Arch Pathol Lab Med. 2011 May;135:558-568.
[8] Bond GL, Levine AJ. A single nucleotide polymorphism in the p53 pathway interacts with gender, environmental stresses and tumor genetics to influence cancer in humans. Onc 2007;26:1317-1323.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
13
Gergelits V. – O vlastnostech genov´e konverze
O vlastnostech genov´ e konverze V´ aclav Gergelits1,2 1 2
´ ˇ Praha, Cesk´ ˇ a republika Ustav molekul´ arn´ı genetiky AV CR,
ˇ a republika 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova v Praze, Praha, Cesk´
Kontakt: V´ aclav Gergelits ´ ˇ v.v.i. Ustav molekul´ arn´ı genetiky AV CR, Adresa: V´ıdeˇ nsk´ a 1083, 142 20 Praha 4 E–mail:
[email protected]
C´ıle v´ yzkumu
Souˇ casn´ y stav pozn´ an´ı
Meiotick´a rekombinace homologn´ıch chromozom˚ u je z´asadn´ı pro zvyˇsov´ an´ı vnitrodruhov´e a mezidruhov´e genetick´e diverzity a spr´ avnou segregaci chromozom˚ u do gamet. Proces meiotick´e rekombinace zaˇc´ın´a v prvn´ı meiotick´e prof´azi, kdy enzym SPO11 vytv´ aˇr´ı na vˇsech ˇctyˇrech chromatid´ ach dvouˇretˇezcov´e zlomy (double-strand break, DSB). Vznik DSB je na jedn´e stranˇe nezbytn´ y pro v´ yvin genetick´e diverzity, na druh´e stranˇe je ale pro pˇreˇzit´ı buˇ nky nutn´e vˇsechny DSB opravit. Proces opravy konkr´etn´ıho DSB zaˇc´ın´ a uvolnˇen´ım jednoho vl´ akna DNA, kter´e si hled´a dostateˇcnˇe podobnou sekvenci u chromatidy homologn´ıho chromozomu. DSB je moˇzn´e opravit i pomoc´ı sestersk´e chromatidy, na autozomech u savc˚ u je vˇsak tato moˇznost dosud jen hypotetick´ a.
Je zn´amo, ˇze m´ısta CO se v genomu nevyskytuj´ı n´ahodnˇe, ale shlukuj´ı se v m´ıstech d´elky v ˇr´adu kilob´az´ı, kde je jejich v´ yskyt v´ yraznˇe pravdˇepodobnˇejˇs´ı, v tzv. rekombinaˇcn´ıch hotspotech. Rekombinaˇcn´ı hotspoty jsou u savc˚ u urˇcov´any t´emˇeˇr v´ yluˇcnˇe genem Prdm9. Kaˇzd´a alelick´a forma genu Prdm9 urˇcuje vazebn´a m´ısta – m´ısta na genomu d´elky v ˇr´adu des´ıtek b´az´ı, na kter´ ych se m˚ uˇzou vytv´aˇret DSB. Novˇe se ukazuje, ˇze v tˇech sam´ ych hotspotech se vyskytuj´ı tak´e NCO [2]. Existuje nˇekolik studi´ı, kter´e se zab´ yvaly odhadem d´elky u ´sek˚ u genov´e konverze u savc˚ u zaloˇzen´ ych na ´ r˚ uzn´ ych typech dat. Useky genov´e konverze doprov´azej´ıc´ı NCO jsou typicky kratˇs´ı neˇz u ´seky doprov´azej´ıc´ı CO. D´elka genov´ ych konverz´ı pˇri NCO se odhadovala na 55– 290 bp [5]. Dvˇe souˇcasn´e pr´ace referuj´ı o detekci u ´sek˚ u genov´e konverze pˇri NCO d´elky 86 ± 49 bp [4] a 100– ´ 1000 bp [2]. Useky genov´e konverze pˇri CO byly poprv´e pˇr´ımo detekov´any tetr´adovou anal´ yzou a jejich d´elka byla urˇcena pro dva r˚ uzn´e hotspoty: 626 ± 319 bp a 566 ± 277 bp [4]. Pˇredmˇetem z´ajmu je t´eˇz pod´ıl v´ yskytu ud´alost´ı NCO a ud´alost´ı CO. Podle poˇctu DSB, kter´e se vyskytnou bˇehem mei´ozy, lze usuzovat, ˇze opravy DSB pomoc´ı NCO probˇehnou v 90 % pˇr´ıpad˚ u a pomoc´ı CO v 10 % pˇr´ıpad˚ u. ˇ Cetnost v´ yskytu NCO a CO se ale napˇr´ıˇc genomem odliˇsuje. Ve dvou r˚ uzn´ ych hotspotech byly pozorov´any pomˇery NCO:CO 1:1 a 15:1 [4]. Podle kvantitativn´ıch vlastnost´ı genov´e konverze lze usuzovat, jak´ ym mechanismem prob´ıhaj´ı. Klasick´ y Szostak˚ uv model [6] pˇredpokl´adal, ˇze u savc˚ u k opravˇe DSB doch´az´ı v obou procesech CO i NCO mechanismem rozruˇsen´ı dvojit´eho Hollydayova spoje (double Hollyday Junction, dHJ). Nyn´ı vˇsak Francesca Cole a kolektiv ve sv´e pr´aci [4] naznaˇcuj´ı, ˇze se jedn´a o jin´ y mechanismus, napˇr. Synthesis-dependent strand annealing (SDSA). Ud´alosti genov´e konverze spolu s CO pˇrisp´ıvaj´ı k evoluci genomu. Mechanismem genov´e konverze se kop´ıruje u ´sek DNA d´elky v ˇr´adu des´ıtek aˇz stovek b´az´ı z d´arcovsk´e chromatidy do pˇrij´ımaj´ıc´ı homologn´ı chroma-
Po u ´spˇeˇsn´em nalezen´ı homologn´ıho u ´seku doch´az´ı k rekombinaci a DSB se oprav´ı. Rekombinace m˚ uˇze probˇehnout dvˇema r˚ uzn´ ymi zp˚ usoby. Bud’ se po kontaktu chromatidov´a ram´enka vymˇen´ı – crossover (CO), nebo se chromatidov´a ram´enka nevymˇen´ı – noncrossover (NCO). Pˇri obou tˇechto zp˚ usobech doch´ az´ı ke genov´e konverzi, jednosmˇern´emu pˇresunu informace z d´ arcovsk´e chromatidy do druh´e, pˇrij´ımaj´ıc´ı chromatidy. O procesu NCO je zn´ amo m´enˇe neˇz o procesu CO. Detekce genov´e konverze je dobr´ ym zp˚ usobem, jak uk´azat na m´ısta na genomu, kde probˇehla NCO, a l´epe tak pochopit mechanismus NCO a genov´e konverze. V´ yzkum genov´e konverze je obecnˇe motivov´ an jej´ı rol´ı v evoluci a dynamice genomu, snahou o popis mechanismu rekombinace a kv˚ uli jej´ımu vztahu k lidsk´ ym nemocem [1]. V m´em doktorsk´em projektu se zab´ yv´ am moˇznostmi detekce element˚ u genov´e konverze pˇri NCO za vyuˇzit´ı myˇs´ıho modelu. C´ılem je detekce a charakterizace element˚ u genov´e konverze v celochromozom´ aln´ım mˇeˇr´ıtku a n´asledn´a inference, jak´ ymi mechanismy genov´a konverze a NCO prob´ıh´ a. Pˇredmˇetem tohoto ˇcl´ anku je pojedn´an´ı o z´akladn´ıch vlastnostech genov´e konverze a NCO u ˇclovˇeka a savc˚ u obecnˇe vˇcetnˇe jej´ıho vztahu k lidsk´ ym nemocem. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
14
Gergelits V. – O vlastnostech genov´e konverze
tidy. D´arcovsk´a chromatida z˚ ust´ av´ a nezmˇenˇena, zat´ımco v´ ysledkem je substituce, inzerce, nebo delece sekv pˇrij´ımaj´ıc´ı chromatidˇe je nahrazen p˚ uvodn´ı u ´sek DNA vence DNA u pˇr´ıjemce. d´arcovsk´ ym u ´sekem. Genov´a konverze se na evoluci genomu podepisuje Zdroj: [3] nˇekolika zp˚ usoby; uved’me tˇri pˇr´ıklady: SNOMED CT: nenalezeno 1. Genov´a konverze zp˚ usobuje GC vych´ ylen´ı – ˇcastˇeji totiˇz doch´az´ı k jednostrann´emu pˇrenosu polymor- MeSH: D005785 fism˚ u s b´azemi G a C neˇz s b´ azemi A a T. Uvaˇzuje ICD10: nenalezeno se, ˇze se toto GC vych´ ylen´ı kompenzuje n´ ahradou nukleotidu C nukleotidem T, ke kter´e doch´ az´ı v souHomologn´ı rekombinace vislosti s pˇredchoz´ı metylac´ı cytosinu. ymˇenˇe dvou 2. Genov´a konverze velkou mˇerou zapˇr´ıˇciˇ nuje evo- Definice: Proces, pˇri kter´em doch´az´ı k v´ u ´sek˚ u DNA mezi dvˇema duplexy DNA, kter´e sd´ılej´ı luci centromer, kter´e jinak podstupuj´ı v´ yraznˇe niˇzˇs´ı vysokou m´ıru podobnosti. poˇcet CO neˇz okoln´ı oblasti [7]. 3. Genov´a konverze souvis´ı s tzv. hotspotov´ ym paradoxem: Jak m˚ uˇzou rekombinaˇcn´ı hotspoty, kter´e jsou urˇcov´any in-trans genem Prdm9, v˚ ubec existovat? Pr´avˇe d´ıky tomu, ˇze to jsou rekombinaˇcn´ı hotspoty, se na jejich m´ıstˇe ˇcasto utv´ aˇr´ı DSB, kter´e jsou opravov´any procesem, pˇri kter´em doch´ az´ı ke genov´e konverzi. Opravou DSB doch´ az´ı ale k modifikaci vazebn´eho m´ısta, kter´e pak uˇz d´ ale nem˚ uˇze b´ yt vazebn´ ym m´ıstem. Pr´ ace F. Cole a koleg˚ u [4] pˇrinesla ˇc´asteˇcn´e vysvˇetlen´ı tohoto paradoxu. Ukazuje se v n´ı, ˇze k erozi vazebn´ ych m´ıst doch´ az´ı zhruba jen v 20 % pˇr´ıpad˚ u genov´e konverze pˇri NCO. Ve zbytku pˇr´ıpad˚ u je genovou konverz´ı zmˇenˇeno pouze okol´ı vazebn´eho m´ısta.
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı
Zdroj: [1] SNOMED CT: nenalezeno MeSH: D059765 ICD10: nenalezeno
Dvouˇretˇ ezcov´ y zlom Definice: Zlomy v protilehl´ ych vl´aknech DNA, jejichˇz vz´ajemn´a vzd´alenost je 10–20bp. Synonyma: DSB Zdroj: [1] SNOMED CT: nenalezeno MeSH: D053903
NCO a genov´a konverze jsou jedn´ım z proces˚ u, ke ICD10: nenalezeno kter´ ym hojnˇe doch´az´ı bˇehem mei´ ozy, bunˇeˇcn´eho dˇelen´ı, kter´e prob´ıh´a u vˇsech pohlavnˇe se rozmnoˇzuj´ıc´ıch euka- Reference ryot. NCO tedy pˇrirozenˇe souvis´ı i s patogenez´ı u ˇclovˇeka. [1] Chen, J. M., Cooper, D. N., Chuzhanova, N., F´ erec, C., & PaExistuje minim´alnˇe 18 lidsk´ ych nemoc´ı, jejichˇz vznik soutrinos, G. P. (2007). Gene conversion: mechanisms, evolution vis´ı s projevy genov´e konverze, uved’me napˇr.: Kampoand human disease. Nature Reviews Genetics, 8(10), 762-775. melick´a dysplazie, Hypergonadotrofick´ y hypogonadismus, [2] Williams, A. L., Genovese, G., Dyer, T., Altemose, N., Truax, syndrom Hurler/Scheie [1]. Na z´ akladˇe naˇsich nepublikoK., Jun, G., & Przeworski, M. (2015). Non-crossover gene conversions show strong GC bias and unexpected clustering in huvan´ ych dat se domn´ıv´ame, ˇze genov´ a konverze m˚ uˇze soumans. eLife, 4, e04637. viset s mechanismy hybridn´ı sterility.
Podˇ ekov´ an´ı
[3] Assis, R., & Kondrashov, A. S. (2012). A strong deletion bias in nonallelic gene conversion. PLoS Genet, 8(2), e1002508e1002508.
Studie byla podpoˇrena projektem SVV 260158 Univerzity Karlovy v Praze.
[4] Cole, F., Baudat, F., Grey, C., Keeney, S., de Massy, B., & Jasin, M. (2014). Mouse tetrad analysis provides insights into recombination mechanisms and hotspot evolutionary dynamics. Nature genetics, 46(10), 1072-1080.
Kl´ıˇ cov´ a slova
[5] Jeffreys, A. J., & May, C. A. (2004). Intense and highly localized gene conversion activity in human meiotic crossover hot spots. Nature genetics, 36(2), 151-156.
Genov´ a konverze
[6] Szostak, J. W., Orr-Weaver, T. L., Rothstein, R. J., & Stahl, F. W. (1983). The double-strand-break repair model for recombination. Cell, 33(1), 25-35.
Definice: Genov´a konverze je proces, pˇri kter´em je jedna sekvence DNA kop´ırov´ ana z jednoho u ´seku genomu (d´arcovsk´eho) do druh´eho (pˇrij´ımaj´ıc´ıho) a jehoˇz
[7] Shi, J., Wolf, S. E., Burke, J. M., Presting, G. G., Ross-Ibarra, J., & Dawe, R. K. (2010). Widespread gene conversion in centromere cores. PLoS Biol, 8(3), e1000327.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
15
Hruˇskov´a L. – Detekce mutac´ı v genech kolagenu typ I
Detekce mutac´ı v genech kolagenu typ I Lucie Hruˇskov´ a1 1
ˇ a republika Klinika dˇetsk´eho a dorostov´eho l´ekaˇrstv´ı, 1. l´ekaˇrsk´ a fakulta, Karlova univerzita v Praze, Cesk´
Kontakt: Lucie Hruˇskov´ a 1. l´ ekaˇrsk´ a fakulta, Karlova univerzita v Praze Adresa: Kateˇrinsk´ a 32, 121 08 Praha 2 E–mail:
[email protected]
C´ıle v´ yzkumu Kolagen typ I je hlavn´ım struktur´ aln´ım proteinem pojivov´e tk´anˇe, zejm´ena kost´ı a k˚ uˇze. Jedn´ a se o heterotrimer skl´adaj´ıc´ı se ze dvou alpha1 (I) ˇretˇezc˚ u a jednoho alpha2 (I) ˇretˇezce, kter´e jsou k´ odov´ any geny COL1A1 a COL1A2. Kolagen typ I vytv´ aˇr´ı vazby s dalˇs´ımi molekulami extracelul´arn´ı matrix (cartilage oligomeric matrix protein, integriny, decorin, kolagen typ V, phosphophoryn a dalˇs´ı), ˇc´ımˇz zvyˇsuje integritu pojivov´e tk´ anˇe [1, 2]. C´ılem tohoto v´ yzkumu je kompletace molekul´arnˇe genetick´ ych dat pacient˚ u s postiˇzenou tvorbou kolagenu typ I a porovn´an´ı z´ıskan´ ych dat s fenotypem tˇechto jedinc˚ u. Studie je zamˇeˇrena na anal´ yzu DNA vzork˚ u (s vyuˇzit´ım technik polymer´azov´e ˇretˇezov´e reakce (PCR) a Sangerova sekvenov´an´ı) pacient˚ u s diagn´ ozou osteogenesis imperfecta, typ I–IV.
Vznikl´ y transkript nen´ı inkorporov´an do molekul kolagenu typ I, coˇz m´a za n´asledek sn´ıˇzenou produkci proteinu. OI typy II–IV maj´ı sv˚ uj p˚ uvod ve struktur´aln´ıch zmˇen´ach gen˚ u COL1A1 a COL1A2. 80% z nich je zp˚ usobeno substitucemi glycinu (nejd˚ uleˇzitˇejˇs´ı aminokyseliny alpha ˇretˇezc˚ u, v jejichˇz sekvenci se vyskytuje v 338 Gly-X-Y repetic´ıch), 20% je v´ ysledkem mutac´ı v oblastech sestˇrihu genu. Z´avaˇznost mutac´ı stoup´a smˇerem k Ckonci genu, nebot’ skl´ad´an´ı alpha ˇretˇezc˚ u do heterotrimeru zaˇc´ın´a pr´avˇe v jejich C-koncov´e oblasti [3]. V r´amci molekuly kolagenu typ I jsou vyˇclenˇeny tˇri oblasti (tzv. Multi ligand binding regions) s vysokou koncentrac´ı mezimolekul´arn´ıch vazeb. Mutace v prvn´ı z tˇechto oblast´ı (MLBR1) maj´ı obvykle za n´asledek m´ırn´ y fenotypov´ y projev (OI typy I a IV), zat´ımco strukturn´ı zmˇeny MLBR2 a MLBR3 zp˚ usobuj´ı v´aˇzn´e a let´aln´ı formy onemocnˇen´ı (OI typy II a III) [4].
Souˇ casn´ y stav pozn´ an´ı Genetick´e zmˇeny kolagenu typ I maj´ı za n´asledek Uplatnˇ en´ı v biomedic´ınˇ e bud’ produkci defektn´ıch molekul proteinu, nebo sn´ıˇzenou a zdravotnictv´ı synt´ezu koneˇcn´eho produktu. Mutace gen˚ u COL1A1 a COL1A2 jsou spojov´ any s onemocnˇen´ımi, jako jsou osteogenesis imperfecta typ I–IV, Ehlers˚ uv - Danlos˚ uv synMolekul´arnˇe-genetick´e anal´ yzy gen˚ u kolagenu typ drom (Klasick´ y typ a typ VIIA), Caffeyova nemoc a idioI jsou d˚ u leˇ z it´ e pro identifikaci co moˇ zn´a nejvˇetˇs´ıho patick´a osteopor´oza [2]. mutaˇcn´ıho spektra v r´amci ˇcesk´ ych OI pacient˚ u. Z´ıskan´a Osteogenesis imperfecta (OI), typ I–IV, je dˇediˇcn´e one- data budou porovn´av´ana se svˇetov´ ymi datab´azemi (Onmocnˇen´ı kost´ı projevuj´ıc´ı se fragiln´ı kostn´ı tk´ an´ı, menˇs´ım line Mendelian Inheritance in Man, Human Genome Muvzr˚ ustem pacient˚ u, deformacemi kost´ı, modr´ ym ˇci ˇsed´ ym tation Database, Ensembl, GeneCards, atd.) za u ´ˇcelem zbarven´ım skl´er, poruchou tvorby dentinu (dentinogene- srovn´an´ı etiologie tohoto onemocnˇen´ı u ˇcesk´ ych pacisis imperfecta) a dalˇs´ımi. V´ yskyt onemocnˇen´ı je uv´adˇen ent˚ u s dalˇs´ımi populacemi ˇci etniky. Identifikace DNA 1 : 15000–20000 ˇzivˇe narozen´ ych. Z´ avaˇznost onemocnˇen´ı je zmˇen a jejich klinick´eho projevu m˚ uˇze b´ yt n´apomocn´e m´ırn´a aˇz let´aln´ı, pˇriˇcemˇz pˇr´ıtomnost klinick´ ych znak˚ u se pro zah´ajen´ı vˇcasn´e a vhodn´e l´eˇcby pacienta. Z´aroveˇ n liˇs´ı nejen mezi jednotliv´ ymi typy onemocnˇen´ı, ale z´aroveˇ n m˚ uˇze b´ yt uˇziteˇcn´ ym krokem pro predikˇcn´ı diagnostiku jev r´amci pacient˚ u jedn´e formy OI [3]. dinc˚ u se suspektn´ım defektem kolagenu typ I. V pˇr´ıpadˇe Prvn´ı typ OI je ve vˇetˇsinˇe pˇr´ıpad˚ u zp˚ usoben vzni- negativn´ıch v´ ysledk˚ u mutaˇcn´ı anal´ yzy gen˚ u COL1A1 kem tzv. STOP kodonu, jehoˇz pˇr´ıtomnost v nukleotidov´e a COL1A2 budou k doplˇ nuj´ıc´ım anal´ yz´am vybr´any dalˇs´ı sekvenci genu vede k pˇredˇcasn´emu ukonˇcen´ı transkripce. geny asociovan´e s geny kolagenu typ I. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
16
Hruˇskov´a L. – Detekce mutac´ı v genech kolagenu typ I
Podˇ ekov´ an´ı
Mutace
´rovni Tato pr´ace byla podpoˇrena projektem SVV-2015- Definice: Zmˇena genetick´e dˇediˇcn´e informace na u DNA t´ ykaj´ıc´ı se bud’ gen˚ u, nebo cel´ ych chro260158 Univerzity Karlovy v Praze. mozom˚ u. Podle m´ısta, kter´e je zasaˇzeno, m˚ uˇze ˇci nemus´ı ovlivˇ novat funkci buˇ nky a organismu. Kl´ıˇ cov´ a slova Vznik´a samovolnˇe v´ yznam ve v´ yvoji druh˚ u, nebo je zp˚ usobena zevn´ımi faktory – mutageny chemick´e, fyAminokyselina zik´aln´ı nebo biologick´e vlivy. Z´avaˇzn´a m. m˚ uˇze v´est k z´aniku buˇ nky, k poruˇse jej´ı funkce a nˇekdy t´eˇz Definice: Organick´a molekula obsahuj´ıc´ı aminoskupinu k jej´ımu zhoubn´emu bujen´ı, viz kancerogeny. M. zai karboxyl. ?-Aminokyseliny (ve kter´ ych jsou amisahuj´ıc´ı pohlavn´ı buˇ nku m˚ uˇze b´ yt pˇrenesena na ponoskupina a karboxyl pˇripojeny ke stejn´emu uhl´ıku) tomstvo. Mutac´ı urˇcit´eho genu je zp˚ usobena ˇrada jsou stavebn´ımi kameny protein˚ u. dˇediˇcn´ ych nemoc´ı, napˇr. albinismus, fenylketonurie, hemofilie aj. Pˇred´av´an´ı takto mutovan´eho genu ´ Zdroj: Alberts B a kol. Z´ aklady bunˇeˇcn´e biologie. Uvod se ˇr´ıd´ı pˇr´ısluˇsn´ ymi pravidly dˇediˇcnosti, srov. domi´ ı nad Labem: do molekul´arn´ı biologie buˇ nky. Ust´ nantn´ı, recesivn´ı, Mendelova pravidla. Nov´e metody Espero Publishing; 1998. str. G-2. molekul´arn´ı biologie jsou v´ yznamn´e v diagn´oze m. a v budoucnu se mohou uplatˇ novat i pˇri jejich l´eˇcbˇe, SNOMED CT: 52518006 srov. genov´a terapie. MeSH: D000596
Zdroj: http://lekarske.slovniky.cz/pojem/mutace
ICD10: nenalezeno
SNOMED CT: 55446002 (ID nalezeno pro ”Genetic mutation”)
Deoxyribonukleov´ a kyselina
MeSH: D009154 ICD10: nenalezeno
Definice: Druh nukleov´e kyseliny, kter´ a je z´ akladem dˇediˇcn´e informace, viz gen, chromozom. U eukaryont˚ u je uloˇzena pˇrev´ aˇznˇe v j´ adˇre buˇ nky mal´e mnoˇzstv´ı t´eˇz v mitochondri´ıch. Je tvoˇrena dvˇema dlouh´ ymi ˇretˇezci navz´ ajem spir´ alovitˇe obtoˇcen´ ymi, jejichˇz z´akladn´ımi stavebn´ımi kameny jsou nukleotidy liˇs´ıc´ı se pˇr´ıtomnost´ı ˇctyˇr r˚ uzn´ ych b´ az´ı, jejichˇz jedineˇcn´e seskupen´ı v ˇretˇezci je podkladem informace v DNA uloˇzen´e, viz genetick´ y k´ od. Jde o adenin A, guanin G, cytosin C a thymin T. Oba ˇretˇezce jsou navz´ ajem doplˇ nkov´e – komplement´arn´ı na z´akladˇe zcela pˇresn´eho p´ arov´ an´ı b´ az´ı mezi ˇretˇezci pomoc´ı vod´ıkov´ ych m˚ ustk˚ u A tvoˇr´ı p´ ar s T, G s C, a tak po rozvinut´ı je ke kaˇzd´emu ze samostatn´ ych ˇretˇezc˚ u moˇzn´e utvoˇrit ˇretˇezec nov´ y, zcela totoˇzn´ y s p˚ uvodn´ım ˇretˇezcem replikace. To je z´aklad mnoˇzen´ı a zachov´ an´ı dˇediˇcn´e informace. Z´ aroveˇ n se podle tohoto vzoru tvoˇr´ı molekuly RNA, kter´e se pod´ılej´ı na pˇrenosu a dalˇs´ım zpracov´ an´ı t´eto uloˇzen´e informace. Synonyma: DNA, DNK Zdroj: http://www.lekarske.slovniky.cz/pojem/ dna SNOMED CT: 24851008
Nesmysln´ y kodon Definice: Jeden ze tˇr´ı triplet˚ u (UAG, UAA, UGA), kter´e zp˚ usob´ı ukonˇcen´ı proteosynt´ezy. Synonyma: STOP kodon Zdroj: http://user.mendelu.cz/urban/vsg1/ geneticky_slovnik4_i.htm SNOMED CT: nenalezeno MeSH: D018389 ICD10: nenalezeno
Polymer´ azov´ a ˇretˇ ezov´ a reakce Definice: Technika slouˇz´ıc´ı k namnoˇzen´ı urˇcit´ ych u ´sek˚ u DNA mnohokr´at opakovan´ ym cyklem polymerace DNA a kr´atk´eho zahˇr´at´ı, pˇri nˇemˇz se oddˇel´ı separovan´a komplement´arn´ı vl´akna. Synonyma: PCR ´ Zdroj: Alberts B a kol. Z´aklady bunˇeˇcn´e biologie. Uvod ´ do molekul´arn´ı biologie buˇ nky. Ust´ı nad Labem: Espero Publishing; 1998. str. G-13 SNOMED CT: 702675006
MeSH: D004247 ICD10: nenalezeno
MeSH: D016133 ICD10: nenalezeno S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
17
Hruˇskov´a L. – Detekce mutac´ı v genech kolagenu typ I
Reference [1] Barnes MA, Weizhong Ch, Morello R, Cabral WA, Weis M, Eyre DR et al. Deficiency of cartilage-associated protein in recessive lethal osteogenesis imperfecta. N Engl J Med. 2006; 355(26):2757-2764. [2] Fahiminiya S, Majewski J, Mort J, Moffatt P, Glorieux FH, Rauch F. Mutations in WNT1 are a cause of osteogenesis imperfecta. J. Med. Genet. 2013; 50:345-348.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
[3] Forlino A, Cabral WA, Barnes AV, Marini JC. 2011. New perspectives on osteogenesis imperfecta. Nat Rev Endocrinol. 2011; 7:540–557. [4] Sweeney, SM, Orgel JP, Fertala A, McAuliffe JD, Turner KR, Di Lullo GA, Chen S, Antipova O, Perumal S, Ala-Kokko L, Forlino A, Cabral WA, Barnes AM, Marini JC, San Antonio JD. Candidate cell and matrix interaction domains on the collagen fibril, the predominant protein of vertebrates. J Biol Chem. 2008;283:21187-21197.
18
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
Statistick´ e metody pro tvorbu ˇ casovˇ e z´ avisl´ ych percentilov´ ych graf˚ u pro hodnocen´ı velikosti plodu a dataci gravidity na z´ akladˇ e longitudin´ aln´ıch dat Martin Hynek1,2 , Jeffrey D. Long3,4 , David Stejskal2 , Jana Zv´ arov´ a2,5 1 2 3 4 5
Gennet, Centrum pro fet´ aln´ı medic´ınu a reprodukˇcn´ı genetiku, Praha
´ Ustav hygieny a epidemiologie, 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova, Praha
Psychiatrick´e oddˇelen´ı, Carverova l´ekaˇrsk´ a fakulta, Univerzita v Iowˇe, Iowa City, IA, USA Oddˇelen´ı pro biostatistiku, Fakulta veˇrejn´eho zdrav´ı, Univerzita v Iowˇe, Iowa City, IA, USA
´ ˇ Praha EuroMISE, Evropsk´e centrum pro medic´ınskou informatiku, statistiku a epidemiologii, Ustav informatiky Akademie vˇed CR,
Kontakt: Martin Hynek Gennet, Centrum pro fet´ aln´ı medic´ınu a reprodukˇ cn´ı genetiku Adresa: Kosteln´ı 9, 170 00 Praha 7 E–mail:
[email protected]
´ Uvod Hodnocen´ı velikosti plodu a urˇcen´ı pˇresn´eho gestaˇcn´ıho st´aˇr´ı (GS) maj´ı z´ asadn´ı v´ yznam pro spr´avn´ y management tˇehotenstv´ı. Vˇcasn´e odhalen´ı restrikce r˚ ustu nebo makrosomie u plodu m˚ uˇze sn´ıˇzit s t´ım spojenou morbiditu a mortalitu [1, 2]. Znalost pˇresn´eho GS m˚ uˇze zabr´anit zbyteˇcn´ ym peripart´ aln´ım intervenc´ım [3]. Jak hodnocen´ı velikosti, tak datace gravidity se t´emˇeˇr bezv´ yhradnˇe op´ıraj´ı o ultrazvukov´e mˇeˇren´ı fet´ aln´ıch biometrick´ ych parametr˚ u (napˇr. temeno-kostrˇcn´ı d´elka, obvod hlaviˇcky, d´elka stehenn´ı kosti apod.), z nichˇz t´emˇeˇr vˇsechny v pr˚ ubˇehu gravidity rostou. K jejich posouzen´ı slouˇz´ı ˇcasovˇe z´avisl´e percentilov´e grafy (referenˇcn´ı grafy), pomoc´ı kter´ ych porovn´ ame z´ıskan´ y biometrick´ y parametr s oˇcek´avanou hodnotou dan´eho parametru v referenˇcn´ı populaci. V literatuˇre lze nal´ezt celou ˇradu postup˚ u vhodn´ ych pro konstrukci percentilov´ ych graf˚ u. Necht’ y pˇredstavuje fet´aln´ı biometrick´ y parametr a t gestaˇcn´ı st´ aˇr´ı. Statistick´ ym c´ılem je odhadnout kvantil, cα (y|t), podm´ınˇen´eho rozdˇelen´ı y (pro dan´e t) pro stanoven´e hodnoty α (napˇr. α = 0, 95 pro (100 · α) = 95. percentil). Kromˇe toho poˇzadujeme, aby cα (y|t) byl hladkou funkc´ı t [4]. Pˇri odhadu cα (y|t) nar´aˇz´ıme na dva hlavn´ı probl´emy: aproximace distribuce y|t a vyhlazen´ı odhadnut´ ych kvantil˚ u v z´avislosti na t. K ˇreˇsen´ı tˇechto probl´em˚ u bylo navrˇzeno nˇekolik moˇzn´ ych pˇr´ıstup˚ u, vˇcetnˇe parametrick´ ych, semiparametrick´ ych a neparametrick´ ych metod. Podrobn´e
pˇrehledy a srovn´an´ı r˚ uzn´ ych pˇr´ıstup˚ u lze nal´ezt v literatuˇre [5, 6, 7, 8]. Naˇs´ım z´ajmem m˚ uˇze b´ yt vytvoˇren´ı dvou typ˚ u referenˇcn´ıch graf˚ u: grafy pro hodnocen´ı velikosti plodu a dataˇcn´ı kˇrivky. Prvn´ı jmenovan´e slouˇz´ı k hodnocen´ı velikosti plodu, zat´ımco ty druh´e k odhadu GS. Je nespr´avn´e pouˇz´ıvat velikostn´ı percentilov´e grafy ke stanoven´ı GS. K tomuto u ´ˇcelu je potˇreba c´ılenˇe vytvoˇrit dataˇcn´ı grafy [9]. Jestliˇze velikostn´ı grafy modeluj´ı velikost plodu jako funkci GS, y|t, potom dataˇcn´ı grafy konstruujeme analogicky jako velikostn´ı grafy s t´ım rozd´ılem, ˇze GS modelujeme jako funkci velikosti plodu, t|y. Metoda vhodn´a pro konstrukci percentilov´ ych graf˚ u by mˇela splˇ novat urˇcit´e poˇzadavky. V roce 1993 uvedl Altman a Chittyov´a, ˇze percentilov´e kˇrivky by se mˇely hladce mˇenit s pr˚ ubˇehem gestace a vykazovat dobrou shodu s daty. D´ale je ˇz´adouc´ı, aby byl statistick´ y model tak jednoduch´ y, jak je to moˇzn´e v souladu s tˇemito poˇzadavky [9]. Experti studijn´ı komise Svˇetov´e zdravotnick´e organice (WHO) pro tvorbu nov´ ych r˚ ustov´ ych graf˚ u u dˇet´ı (Multicentre Growth Reference Study, MGRS) [8] se shodli na tom, ˇze mezi prim´arn´ı krit´eria kladen´a na metody pro tvorbu percentilov´ ych graf˚ u patˇr´ı schopnost: • pˇresnˇe odhadnout zevn´ı percentily, • odhadnout vˇsechny percentily souˇcasnˇe takov´ ym zp˚ usobem, aby se nemohly kˇr´ıˇzit, • urˇcit Z-sk´ore a percentily pˇr´ımo pomoc´ı vzorc˚ u, • kontinu´alnˇe vyhladit kˇrivky v z´avislosti na vˇeku a S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
19
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
• v pˇr´ıpadˇe potˇreby poˇc´ıtat s ˇsikmost´ı a ˇspiˇcatost´ı. Kromˇe toho jako sekund´ arn´ı kriteria urˇcili, aby metoda • umoˇzn ˇovala posoudit shodu modelu s daty, • byla jednoduˇse vysvˇetliteln´ a a dobˇre dokumentovateln´a, • byla aplikovateln´ a na r˚ uzn´e antropometrick´e parametry, aby bylo moˇzn´e vytvoˇrit WHO kˇrivky na z´akladˇe jednotn´eho pˇr´ıstupu. Vˇetˇsina literatury, kter´ a se zab´ yv´ a tvorbou percentilov´ ych graf˚ u, popisuje statistick´e metody aplikovateln´e na data z pr˚ uˇrezov´ ych studi´ı, tj. kdy kaˇzd´ y plod pˇrisp´ıv´a do reprezentativn´ıho v´ ybˇeru pouze jedn´ım mˇeˇren´ım. Protoˇze je sbˇer takov´ ych dat snadnˇejˇs´ı a metodika jejich zpracov´an´ı jednoduˇsˇs´ı, vˇetˇsina studi´ı, kter´e byly publikov´any v posledn´ıch desetilet´ıch, pouˇzila ke konstrukci referenˇcn´ıch graf˚ u data z pr˚ uˇrezov´ ych studi´ı. Nicm´enˇe, v souˇcasn´e dobˇe m´ ame k dispozici st´ ale ˇcastˇeji data z longitudin´aln´ıch studi´ı, tj. takov´ ych, kdy je kaˇzd´ y plod zmˇeˇren v´ıce neˇz jednou, a takov´ a data potom vyˇzaduj´ı odliˇsnou statickou metodiku. Mezi hlavn´ı odliˇsnosti patˇr´ı, ˇze s´eriov´a mˇeˇren´ı jednoho plodu jsou korelov´ ana a s t´ım mus´ı pˇr´ısluˇsn´a metoda poˇc´ıtat. Kromˇe toho je ot´azkou, jak u takov´eho souboru poˇc´ıtat poˇcet stupˇ n˚ u volnosti (df ), a to pˇredevˇs´ım v pˇr´ıpadˇe, kdy se poˇcet opakovan´ ych mˇeˇren´ı mezi plody liˇs´ı [9]. C´ılem tohoto ˇcl´ anku je sestavit a navrhnout statistickou metodiku vhodnou pro tvorbu ˇcasovˇe z´ avisl´ ych referenˇcn´ıch graf˚ u plodu na z´ akladˇe longitudin´aln´ıch dat. Tato metoda bude v budoucnosti pouˇzita pro tvorbu referenˇcn´ıch graf˚ u pro hodnocen´ı velikosti plodu a dataci gravidity pro ˇceskou populaci, a to na z´ akladˇe rozs´ahl´eho souboru shrom´aˇzdˇen´eho bˇehem rutinn´ıch ultrazvukov´ ych vyˇsetˇren´ı v Centru fet´ aln´ı medic´ıny Gennet v Praze. Protoˇze je podstatn´ a ˇc´ ast plod˚ u v naˇsem centru vyˇsetˇrena bˇehem tˇehotenstv´ı v´ıce neˇz jednou, zamˇeˇr´ıme se v tomto ˇcl´anku na adaptaci st´ avaj´ıc´ıch metod pro tvorbu fet´aln´ıch referenˇcn´ıch graf˚ u na data tohoto typu. Tento ˇcl´anek se zamˇeˇruje v´ yhradnˇe na statistickou metodiku. Mezi nem´enˇe d˚ uleˇzit´e body patˇr´ı problematika velikosti a v´ ybˇeru vzorku, kriteri´ı pro zaˇrazen´ı nebo nezaˇrazen´ı pˇr´ıpadu do souboru a sbˇeru dat. Nicm´enˇe, tyto body nejsou obsahem tohoto sdˇelen´ı a v jejich pˇr´ıpadˇe odkazujeme na pˇr´ısluˇsnou literaturu [9, 10, 11, 12].
Statistick´ e metody Pˇri hled´an´ı vhodn´e statistick´e metody, kter´a by splˇ novala v´ yˇse zm´ınˇen´ a krit´eria, si mus´ıme uvˇedomit dvˇe specifika fet´aln´ı biometrie. Za prv´e, je dobˇre zn´amo, ˇze u plodu pr˚ umˇer biometrick´ ych parametr˚ u bˇehem tˇehotenstv´ı monot´ onnˇe roste. Z´ aroveˇ n plat´ı, ˇze variabilita mˇeˇren´ı mezi jednotliv´ ymi plody, kterou lze vyj´adˇrit pomoc´ı smˇerodatn´e odchylky (SD), m´ a tak´e tendenci r˚ ust (s ˇcasem se rozˇsiˇruje). S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Je proto kl´ıˇcov´e pˇri tvorbˇe modelu poˇc´ıtat nejenom se z´avislost´ı pr˚ umˇeru na GS, ale tak´e se z´avislost´ı SD na GS [9]. Za druh´e, z mnoha pˇredchoz´ıch prac´ı v´ıme, ˇze rozdˇelen´ı fet´aln´ıch biometrick´ ych parametr˚ u je bl´ızk´e norm´aln´ımu rozdˇelen´ı pro jak´ekoli GS [12]. Kromˇe toho pr˚ umˇer vˇetˇsiny fet´aln´ı rozmˇer˚ u v z´avislosti na GS roste neline´arnˇe a stejnˇe tak i SD. Proto je nejpouˇz´ıvanˇejˇs´ı metodou pro konstrukci fet´aln´ıch referenˇcn´ıch graf˚ u line´arn´ı regrese s vyuˇzit´ım zlomkov´ ych polynom˚ u (fractional polynomials, FPs) ([13]), pomoc´ı kter´ ych modelujeme zvl´aˇst’ z´avislost pr˚ umˇeru a zvl´aˇst’ SD na GS a pˇredpokl´ad´ame pro kaˇzd´e GS norm´aln´ı rozdˇelen´ı mˇeˇren´ ych parametr˚ u, ale neline´arn´ı z´avislost na GS [4, 13, 14, 15]. Nejprve pop´ıˇseme metodu pr˚ umˇeru a SD s vyuˇzit´ım klasick´ ych polynom˚ u (conventional polynomials, CPs) pro data z pr˚ uˇrezov´ ych studi´ı, n´aslednˇe pˇredstav´ıme jej´ı modifikaci s vyuˇzit´ım FPs a uvedeme metody pro kontrolu shody modelu s daty. Na z´avˇer navrhneme, jak metodiku rozˇs´ıˇrit na longitudin´aln´ı studie.
Pr˚ uˇrezov´ e studie Metoda pr˚ umˇ eru a smˇ erodatn´ e odchylky P˚ uvodn´ı metoda pr˚ umˇeru a SD pˇredstavuje parametrick´ y pˇr´ıstup navrˇzen´ y Altmanem [14] a Roystonem a Wrightem [15]. Metoda vych´az´ı z pˇredpokladu, ˇze v kaˇzd´em GS sleduj´ı namˇeˇren´e hodnoty norm´aln´ı rozdˇelen´ı a modeluje kˇrivky pro pr˚ umˇer a SD v z´avislosti na GS jako polynomick´e funkce. Poˇzadovanou kvantilovou kˇrivku lze potom z´ıskat pomoc´ı vztahu cα = µ + kσ,
(1)
kde k je odpov´ıdaj´ıc´ı kvantil standardizovan´eho norm´aln´ıho rozdˇelen´ı a µ a σ pˇredstavuj´ı pr˚ umˇer a SD pro dan´e GS v referenˇcn´ı populaci. P˚ uvodn´ı metoda sest´av´a z nˇekolika krok˚ u. Nejprve pomoc´ı CPs modelujeme pr˚ umˇer v z´avislosti na GS. Autoˇri rad´ı zaˇc´ıt s kubick´ ym polynomem a postupnˇe sniˇzovat stupeˇ n polynomu, v pˇr´ıpadˇe ˇze se koeficient nejvyˇsˇs´ıho ˇclenu polynomu neliˇs´ı statisticky v´ yznamnˇe od nuly. Jakmile vybereme vhodn´ y model pro pr˚ umˇer, pokraˇcujeme modelac´ı variability okolo pr˚ umˇeru. Zde autoˇri metody vych´az´ı z u ´vahy, ˇze pokud vykazuj´ı mˇeˇren´ı ve vˇsech GS norm´aln´ı rozdˇelen´ı, potom by rezidua ze z´ıskan´eho regresn´ıho modelu pro pr˚ umˇer mˇela m´ıt tak´e norm´aln´ı rozdˇelen´ı a tedy absolutn´ı hodnoty tˇechto rezidu´ı polonorm´aln´ı rozdˇelen´ı. Pr˚ umˇer p polonorm´aln´ıho standar2/π. Necht’ jsou modizovan´eho rozdˇelen´ı je roven ” difikovan´apabsolutn´ı rezidua“ absolutn´ı hodnoty rezidu´ı n´asoben´e π/2. Tedy, jestliˇze z´ısk´ame regresn´ı model modifikovan´ ych absolutn´ıch rezidu´ı v z´avislosti na GS, budou hodnoty pˇredpovˇezen´e t´ımto modelem odhadem SD rezidu´ı pro dan´e GS a t´ım i odhadem pro y. Podobnˇe jako pˇri modelaci kˇrivky pro pr˚ umˇer, i zde pouˇz´ıv´ame k modelaci z´avislosti polynomickou regresi.
20
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
Koneˇcnˇe, protoˇze SD v z´ avislosti na GS roste (heteroskedasticita), mˇeli bychom n´ aslednˇe znovu vyhladit kˇrivku pro pr˚ umˇer a SD s pouˇzit´ım v´ aˇzen´e regrese pomoc´ı metody nejmenˇs´ıch ˇctverc˚ u, kdy jako v´ ahy pouˇzijeme pˇrevr´acenou hodnotu druh´e mocniny odhadovan´ ych SD pro dan´e GS [16]. Nicm´enˇe, Altman a Chittyov´ a uv´adˇej´ı, ˇze efekt na v´ yslednou kˇrivku je t´emˇeˇr vˇzdy nev´ yrazn´ y [9]. Zlomkov´ e polynomy CPs, kter´e p˚ uvodn´ı metoda pr˚ umˇeru a SD podle Altmana a Chittyov´e pouˇz´ıv´ a, trp´ı nˇekolika dobˇre zn´am´ ymi omezen´ımi. Polynomy n´ızk´eho ˇr´ adu nab´ızej´ı jen omezenou ˇsk´alu tvar˚ u a polynomy vysok´eho ˇr´ adu nemus´ı dobˇre odpov´ıdat dat˚ um na okraj´ıch pozorovan´eho obdob´ı. Kromˇe toho, polynomy postr´ adaj´ı asymptoty [13]. Proto v roce 1994 pˇredloˇzili Royston a Altman rozˇs´ıˇrenou rodinu polynom˚ u nazvan´ ych zlomkov´e polynomy (fractional polynomials, FPs) [13]. FPs jsou podobn´e CPs v tom, ˇze jejich ˇcleny jsou mocninn´e funkce nez´avisl´e promˇenn´e a souˇcasnˇe CPs jsou speci´ aln´ım pˇr´ıpadem FPs. Liˇs´ı se ale v tom, ˇze exponenty ve ˇclenech FPs mohou nab´ yvat z´aporn´ ych hodnot a zlomk˚ u. Obvykle jsou moˇzn´e exponenty vyb´ır´ any z mal´e, pˇredem stanoven´e mnoˇziny celoˇc´ıseln´ ych a neceloˇc´ıseln´ ych hodnot: S = {−3; −2; −1; −0, 5; 0; 0, 5; 1; 2; 3}, kde 0 pˇredstavuje transformaci pomoc´ı pˇrirozen´eho logaritmu. Prvky mnoˇziny S jsou hlavn´ı hodnoty Tukeyho ˇzebˇr´ıˇcku mocnin“, kter´ y ” je uˇz´ıv´an pro vˇseobecn´e ˇreˇsen´ı probl´em˚ u s prokl´ ad´an´ım kˇrivek [17], je moˇzn´e vyuˇz´ıt i jin´e sady. Mezi atraktivn´ı vlastnosti FPs patˇr´ı parsimonie (m˚ uˇzeme doc´ılit podobn´e shody modelu s daty jako u CPs, ale s m´enˇe ˇcleny polynomu), ˇsirok´a ˇsk´ala tvar˚ u kˇrivek a schopnost aproximovat asymptoty [4, 18]. Obecn´a definice FP m-t´eho ˇr´ adu (FPm) a s exponenty p = (p1 ≤ . . . ≤ pm ) je potom d´ ana vztahem ∗
φ (t; p) = β0 + φm (t; p) =
m X
βj hj (t),
(2)
pro pj = 6 pj−1 pro pj = pj−1
(3)
j=0
kde h0 (t) = 1 a kde p t j hj (t) = hj−1 (t) log t
pro j = 1, . . . , m a t0 ≡ loge (t) [4, 13]. Dalˇs´ım omezen´ım je t > 0, abychom zajistili, ˇze FP transformace bude vˇzdy definov´ana. Napˇr´ıklad FP prvn´ıho ˇr´ adu (FP1) pro p1 = 0 je β0 + β1 log t, FP druh´eho ˇr´ adu (FP2) s p1 = −2 a p2 = 1 je β0 + β1 t−2 + β2 t a FP tˇret´ıho ˇr´ adu (FP3) s p = (0; 2; 2) je β0 + β1 log t + β2 t2 + β3 t2 log t. FPs s m ≤ 2 nab´ızej´ı ˇsirokou ˇsk´ al˚ u neline´ arn´ıch (a line´arn´ıch) kˇrivek. Proto jsou v praxi jen velmi zˇr´ıdka potˇrebn´e FPs ˇr´adu vyˇsˇs´ıho neˇz druh´eho [4]. FP1 funkce jsou vˇzdy monot´onn´ı, zat´ımco FP2 funkce mohou b´ yt monot´onn´ı nebo nemonot´ onn´ı (konvexn´ı/konk´ avn´ı) a jako speci´aln´ı pˇr´ıpad obsahuj´ı kvadratick´ y polynom (p1 = 1,
p2 = 2). R˚ ust plodu je z biologick´e podstaty monot´onn´ı. Zd´a se proto logick´e, ˇze pro tvorbu fet´aln´ıch referenˇcn´ıch graf˚ u by staˇcily FP1 modely. Avˇsak FP2 funkce mohou m´ıt tak´e sv˚ uj v´ yznam, pˇredevˇs´ım v pˇr´ıpadˇe, kdy v´ yvoj sledujeme v pomˇernˇe dlouh´em ˇcasov´em obdob´ı. Vzhledem k tomu, ˇze FP2 kˇrivky jsou v´ yraznˇe bohatˇs´ı, co se t´ yˇce moˇzn´ ych tvar˚ u, obvykle pˇri tvorbˇe fet´aln´ıch referenˇcn´ıch graf˚ u pouˇz´ıv´ame jak FP1, tak i FP2 funkce. Pˇredchoz´ı zkuˇsenosti s FPs n´am tak´e ukazuj´ı, ˇze v mnoha pˇr´ıpadech byly pr´avˇe FP2 modely vybr´any jako ty nejlepˇs´ı pˇri konstrukci fet´aln´ıch percentilov´ ych graf˚ u [19, 20, 21]. Pˇri pouˇzit´ı FPs regrese se nab´ız´ı ot´azka, jak vybereme mezi moˇzn´ ymi kandid´aty ten nejlepˇs´ı model (modely) pro naˇse data. Obecnˇe pˇredpokl´ad´ame, ˇze FP model je typicky odhadov´an pomoc´ı maxim´aln´ı vˇerohodnosti (ML). Kl´ıˇcovou charakteristikou je potom -2 kr´at nejvyˇsˇs´ı hodnota vˇerohodnostn´ı funkce, oznaˇcovan´a jako deviance. Jestliˇze jsou vˇsechny kandid´atn´ı FP modely stejn´eho ˇr´adu (napˇr. FP1), je potom nejlepˇs´ı model jednoduˇse ten, kter´ y m´a nejniˇzˇs´ı devianci [4]. Ale pokud chceme porovn´avat FP modely r˚ uzn´eho ˇr´adu, potˇrebujeme odliˇsn´ y postup. D˚ uvodem je fakt, ˇze deviance bude u komplexnˇejˇs´ıho modelu vˇzdy niˇzˇs´ı, a to i v pˇr´ıpadˇe, ˇze je do modelu pˇrid´an bezcenn´ y prediktor [18]. T´ım p´adem by byl sloˇzitˇejˇs´ı model upˇrednostnˇen pˇred modelem jednoduˇsˇs´ım. Proto, pokud m´ame rozhodovat mezi rozd´ılnˇe sloˇzit´ ymi modely, vych´azej´ı Royston a Sauerbrei [4] z faktu, ˇze rozd´ıl devianc´ı mezi nulov´ ym modelem s [β1 , . . . , βm ]T = 0 a FP modelem m-t´eho ˇr´adu m´a pˇribliˇznˇe χ2 rozdˇelen´ı o 2m stupn´ıch volnosti. Podobnˇe, rozd´ıl devianc´ı mezi FPm a FP(m − 1) modely sleduje pˇribliˇznˇe χ2 rozdˇelen´ı o dvou stupn´ıch volnosti pro pˇr´ıpad nulov´e hypot´ezy, ˇze dodateˇcn´e β v FPm modelu je nula [4]. Pro srovn´an´ı model˚ u m˚ uˇzeme tedy pouˇz´ıt test pomˇerem vˇerohodnost´ı (likelihood ratio test, LRT). P˚ uvodn´ı sekvenˇcn´ı postup navrˇzen´ y Roystonem a Altmanem [13] postupnˇe srovn´aval FP2 model s FP1, pot´e FP1 s line´arn´ım modelem (p1 = 1) a koneˇcnˇe line´arn´ı model s nulov´ ym modelem (β1 = 0). Nicm´enˇe, tento postup m˚ uˇze v´ yraznˇe zv´ yˇsit pravdˇepodobnost chyby I. druhu v pˇr´ıpadˇe, ˇze vliv t jako prediktora je nev´ yznamn´ y a m˚ uˇze nˇekdy d´at nejednoznaˇcn´ y v´ ysledek [4]. Proto Royston a Sauerbrei tento postup nedoporuˇcuj´ı a navrhuj´ı jin´ y, uzavˇren´ y postup zvan´ y RA2“ [4], kter´ y ” zachov´av´a pravdˇepodobnost chyby I. druhu na zvolen´e hladinˇe v´ yznamnosti α (typicky α = 5%). Postup je n´asleduj´ıc´ı: 1. Otestuj nejlepˇs´ı FP2 model proti nulov´emu modelu (vˇsechna βj>0 = 0) na hladinˇe v´ yznamnosti α a se ˇctyˇrmi stupni volnosti. Jestliˇze je test statisticky nev´ yznamn´ y, je vliv t jako prediktora nesignifikantn´ı. V opaˇcn´em pˇr´ıpadˇe pokraˇcuj. 2. Otestuj nejlepˇs´ı FP2 model proti line´arn´ımu modelu (p1 = 1) s pouˇzit´ım tˇrech stupˇ n˚ u volnosti. Jestliˇze test nen´ı statisticky signifikantn´ı, je fin´aln´ı model line´arn´ı. V opaˇcn´em pˇr´ıpadˇe pokraˇcuj. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
3. Otestuj nejlepˇs´ı FP2 model proti nejlepˇs´ımu FP1 Rozˇ s´ıˇren´ı na longitudin´ aln´ı studie modelu s pouˇzit´ım dvou stupˇ n˚ u volnosti. Jestliˇze test nen´ı statisticky signifikantn´ı, je fin´aln´ı model Longitudin´aln´ı soubor z´ısk´ame pozorov´an´ım dan´eho FP1. V opaˇcn´em pˇr´ıpadˇe je fin´ aln´ı model FP2. jedince opakovanˇe v ˇcase [26]. Jak jsme jiˇz zm´ınili, c´ılem naˇs´ı pl´anovan´e studie je vytvoˇrit fet´aln´ı percentilov´e grafy Metoda pr˚ umˇeru a SD s pouˇzit´ım FPs je prov´adˇena pro ˇceskou populaci na z´akladˇe dat z naˇseho centra. analogicky, jak byla v´ yˇse pops´ ana s jedin´ ym rozd´ılem, ˇze Nˇekter´e plody jsou v pr˚ ubˇehu tˇehotenstv´ı vyˇsetˇreny pouze m´ısto CPs pouˇzijeme FPs pro modelaci z´ avislosti pr˚ umˇeru jednou. Avˇsak u znaˇcn´e ˇc´asti plod˚ u prov´ad´ıme mˇeˇren´ı v´ıce a SD na GS, y|t. neˇz jednou (aˇz cca desetkr´at). Zach´azet s takov´ ymto souborem jako s pr˚ uˇrezovou studi´ı by bylo zjevnˇe chybn´e, Hodnocen´ı shody modelu s daty nebot’ s´eriov´a mˇeˇren´ı jednoho plodu jsou v´ yraznˇe korelov´ana a efektivn´ı velikost souboru odpov´ıd´a sp´ıˇse celBylo opakovanˇe publikov´ ano [6, 8, 9, 12], ˇze je naprosto kov´emu poˇctu plod˚ u neˇz celkov´emu poˇctu mˇeˇren´ı. V tanezbytn´e posoudit, jak´ ym zp˚ usobem se fin´ aln´ı model sho- kov´em pˇr´ıpadˇe je nezbytn´ y pˇredpoklad klasick´e regresn´ı duje s daty. K tomuto u ´ˇcelu byla navrˇzena cel´ a ˇrada dia- anal´ yzy o nez´avislosti pozorov´an´ı zcela jistˇe poruˇsen. gnostick´ ych n´astroj˚ u. Zde pˇredkl´ ad´ ame seznam nejˇcastˇeji Jestliˇze je v´ ybˇer plod˚ u n´ahodn´ y, lze pˇredpokl´adat, ˇze zmiˇ novan´ ych metod, vˇcetnˇe tˇech, kter´e doporuˇcuje MGRS mˇeˇren´ı mezi jednotliv´ ymi plody jsou nez´avisl´a, ale mˇeˇren´ı komise WHO [8]: prov´adˇen´a v r´amci jednoho plodu jsou z´avisl´a. Pˇri anal´ yze pr˚ uˇrezov´ ych dat se soustˇred’ujeme na variabilitu a rozd´ıly • vizu´aln´ı zhodnocen´ı tvaru kvantilov´ ych kˇrivek vyne- mezi jednotliv´ ymi subjekty. Ale v pˇr´ıpadˇe anal´ yzy lonsen´ ych do grafu z´ aroveˇ n s daty, gitudin´aln´ıch dat je nutn´e zahrnout i variabilitu v r´amci jednotliv´ ych subjekt˚ u. Aplikovat metody klasick´e regresn´ı • graf empiricky odhadnut´ ych a vyhlazen´ ych kvantil˚ u anal´ yzy na longitudin´aln´ı data je nespr´avn´e, nebot’ nerevynesen´ ych z´ aroveˇ n do jednoho grafu, flektov´an´ı variability v r´amci subjektu vede k vych´ ylen´ ych odhad˚ u m stˇ r edn´ ıch chyb a z toho plynouc´ ıch statistik. Je • srovn´an´ı pozorovan´ ych a oˇcek´ avan´ ych ˇcetnost´ı nad tedy zˇ r ejm´ e , ˇ z e klasick´ a regresn´ ı anal´ y za je pro longitua pod pˇr´ısluˇsn´ ymi kvantily, din´aln´ı data nevhodn´a a je nutn´e nal´ezt adekv´atn´ı alter• bodov´ y diagram distribuce rezidu´ı vyj´ adˇren´ ych for- nativn´ı metodu [18]. mou Z-sk´ore, Line´ arn´ı sm´ıˇsen´ e regresn´ı modely • norm´aln´ı Q-Q diagram rezidu´ı vyj´ adˇren´ ych formou Z-sk´ore, Line´ arn´ı sm´ıˇsen´e regresn´ı modely (linear mixed effect regression, LMER) pˇredstavuj´ı metodu pro anal´ yzu • detrendovan´ y norm´ aln´ı Q-Q diagram (worm plot, v´ıce´ urovˇ nov´ ych dat, mezi kter´a longitudin´aln´ı data patˇr´ı ˇcerv´ı diagram) rezidu´ı vyj´ adˇren´ ych jako Z-sk´ore, [27]. Opakovan´a mˇeˇren´ı na jednom subjektu jsou koretj. do grafu vyn´ aˇs´ıme empirick´e kvantily m´ınus lov´ana. Tento fakt zohledˇ nuj´ı v LMER modelech ˇcleny norm´aln´ı kvantily proti norm´ aln´ım kvantil˚ um, oznaˇcovan´e jako n´ ahodn´e efekty, kter´e odr´aˇzej´ı rozd´ıly mezi kˇrivkami jednotliv´ ych subjekt˚ u. LMER obsahuje • testy normality rezidu´ı vyj´ adˇren´ ych formou Z-sk´ore tak´e pevn´e efekty, kter´e jsou analogick´e koeficient˚ um v klanebo jejich kombinace, napˇr. Q-statistika [22] (komsick´e line´arn´ı regresi. Ty jsou konstantn´ı mezi jednotbinace test˚ u na ˇctyˇri momenty rozdˇelen´ı, modifikolivci a odr´aˇzej´ı zmˇeny na u ´rovni skupin [18]. Kombinace van´ ych test˚ u podle D’Agostina [23] a testu podle pevn´ ych a n´ahodn´ ych efekt˚ u dala metodˇe LMER jm´eno. Shapira a Wilka), N´ahodn´e efekty m˚ uˇzeme tak´e ch´apat jako rozd´ıl • graf, ve kter´em jsou vynesena vyhlazen´a rezidua mezi regresn´ı kˇrivkou dan´eho jednotlivce a regresn´ı jako pln´a ˇc´ ara spoleˇcnˇe s 95% intervalem spo- kˇrivkou cel´e skupiny. Pˇredstav´ıme-li si line´arn´ı z´avislost, uˇzeme m´ıt dva r˚ uzn´e n´ahodn´e efekty, kter´e ale mohou lehlivosti. Tyto grafy pom´ ahaj´ı zachytit oblasti m˚ b´ y t korelov´ a ny: n´ a hodn´ y absolutn´ı ˇclen (rozd´ıl mezi abvysvˇetluj´ıc´ı promˇenn´e, ve kter´ ych se model nesolutn´ ım ˇ c lenem regresn´ ı kˇrivky jednotlivce a skupiny) dostateˇcnˇe shoduje s daty. Royston a Sauerbrei a n´ a hodnou smˇ e rnici (rozd´ ıl mezi sklonem kˇrivky jed[4] prezentovali sv´e zkuˇsenosti s vyhlazen´ım ponotlivce a skupiny). V pˇ r ´ ıpadˇ e fet´aln´ı biometrie se zd´a moc´ı lok´alnˇe line´ arn´ı regrese (univariate runningadekv´ a tn´ ı br´ a t do u ´ vahy oba n´ahodn´e efekty, tj. model line smoother), kter´ y do programu Stata imples n´ a hodn´ y m absolutn´ ım ˇ c lenem a smˇernic´ı. Biologick´a pomentoval Sasieni [24]. V prostˇred´ı R lze podobn´eho vaha r˚ u stu plodu totiˇ z naznaˇ c uje, ˇze se plody nejenom liˇs´ı v´ ysledku dos´ ahnout pomoc´ı Friedmanova supervypoˇ c a ´ teˇ c n´ ı velikost´ ı, se kterou vstupuj´ ı do pozorovac´ıho obhlazovaˇce (supersmoother) [25]. dob´ı (ta je vyj´adˇrena absolutn´ım ˇclenem), ale i rychlost´ı ustu (tu odr´aˇz´ı smˇernice). Tyto zm´ınˇen´e n´ astroje mohou b´ yt pouˇzity na celou r˚ fin´aln´ı kˇrivku po jej´ım vyhlazen´ı v z´ avislosti na GS nebo Mezi v´ yhody LMER patˇr´ı, ˇze vˇsechny subjekty nelok´alnˇe pro dan´e GS. mus´ı b´ yt mˇeˇren´e ve stejn´em ˇcasov´em okamˇziku ani S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
21
22
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
ˇ m´ıt stejn´ y poˇcet mˇeˇren´ı. Casov´ a nerovnomˇernost je pr´avˇe charakteristika naˇseho souboru, nebot’ jednotliv´e plody se liˇs´ı poˇctem proveden´ ych mˇeˇren´ı (od jednoho do cca deseti) a ˇcas (GS) mˇeˇren´ı se mezi jednotliv´ ymi plody tak´e liˇs´ı (nˇekter´e v kratˇs´ıch ˇcasov´ ych intervalech, jin´e v delˇs´ıch ˇcasov´ ych intervalech). V tomto pohledu pˇredstavuje LMER optim´ aln´ı pˇr´ıstup, protoˇze LMER m´a mimo jin´e tyto vlastnosti:
Sloupce matice Zi jsou typicky podmnoˇzinou sloupc˚ u matice Xi , tedy l ≤ m. Pro FP1 je obvykl´e, ˇze l = m = 1 a ˇze jsou specifikov´any n´ahodn´ y absolutn´ı ˇclen a n´ahodn´a smˇernice. Typick´e pˇredpoklady normality a dalˇs´ı pˇredpoklady lze shrnout jako bi ∼ N (0, G), εi ∼ N (0, Ri ),
s
bi ⊥εi ,
(8)
tj. n´ahodn´e sloˇzky εi a n´ahodn´e efekty bi jsou pro dan´eho • LMER se dok´aˇze vypoˇr´ adat s chybˇej´ıc´ımi pozo- jedince vz´ajemnˇe nez´avisl´e. D´ale pˇredpokl´ad´ame, ˇze Ri = rov´an´ımi. Subjekty, pro kter´e je k dispozici alespoˇ n σε2 Ini , kde Ini je jednotkov´a matice o rozmˇerech ni × ni jedno pozorov´an´ı, mohou b´ yt zahrnuty do anal´ yzy. a G je nezn´am´a kovarianˇcn´ı matice pro n´ahodn´e efekty o rozmˇerech (l + 1) × (l + 1) a s nez´aporn´ ymi variancemi • LMER pˇredstavuje metodu velmi flexibiln´ı, pokud na diagon´ale. Pevn´e efekty a jednotliv´e komponenty vaˇ jde o strukturu dat. Casy pozorov´ an´ı se mohou riability jsou typicky odhadov´any pomoc´ı ML nebo momezi jednotliv´ ymi subjekty liˇsit, stejnˇe tak i ˇcasov´a difikovan´e maxim´aln´ı vˇerohodnosti (restricted maximum vzd´alenost mezi r˚ uzn´ ymi pozorov´ an´ımi nemus´ı b´ yt likelihood, REML) [18, 28]. stejn´a. Napˇr´ıklad, jednoduch´ y line´arn´ı LMER model • LMER dok´aˇze kontrolovat poˇcet a charakter ˇclen˚ u s n´ a hodn´ y m absolutn´ ım ˇ c lenem a smˇernic´ı je pro FP1 potˇrebn´ ych pro modelaci z´ avislosti na ˇcase. Lze s p = 1 (l = m = 1) d´ a n jako 1 pouˇz´ıt CPs, FPs a ˇradu dalˇs´ıch transformac´ı (napˇr. spliny).
yij = (β0 + b0i ) + (β1 + b1i )tij + εij .
(9)
Vr´at´ıme-li se k charakteristik´ am LMER, pˇredpokl´adejme, ˇze yij je pozorov´ an´ı i-t´eho objektu Zlomkov´ e polynomy (i = 1, . . . , N ) v j-t´em ˇcasov´em bodˇe (j = 1, . . . , ni ), m je poˇcet prediktor˚ u pevn´ ych efekt˚ u, a l je poˇcet prediktor˚ u n´ahodn´ ych efekt˚ u (u obou nepoˇc´ıt´ ame absolutn´ı Jak jsme v´ yˇse uk´azali, lze FPs pˇr´ımo zaˇclenit ˇclen). Obecn´a definice klasick´eho LMER modelu je potom do LMER. Z´aroveˇ n jsme jiˇz zm´ınili, ˇze charakter d´ana [28, 29] r˚ ustu plodu pˇredpokl´ad´a model s n´ahodn´ ym absolutn´ım ˇclenem a smˇernic´ı. Lze dok´azat, ˇze korelace (kovariyi = Xi β + Zi bi + εi , (4) ance) mezi n´ahodn´ ymi efekty pˇredpokl´ad´a, ˇze variabilita fet´ a ln´ ıch biometrick´ ych parametr˚ u m˚ uˇze v ˇcase (GS) kde yi je ˇcasov´ y vektor i-t´eho jedince o rozmˇerech ni × 1, r˚ u st, coˇ z je vlastnost d˚ u leˇ z it´ a pro n´ a ˇ s model. V pˇr´ıpadˇe Xi je zn´am´a matice pl´ anu pro pevn´e efekty rozmˇer˚ u FP2 model˚ u nebo i model˚ u vyˇ s ˇ s ´ ıch ˇ r a ´ d˚ u je nutn´e se ni ×(m+1), β je vektor o rozmˇerech (m+1)×1 obsahuj´ıc´ı rozhodnout, kolik n´ a hodn´ y ch efekt˚ u bude do modelu pevn´e, ale nezn´am´e regresn´ı koeficienty, Zi je zn´ am´ a mazaˇ c lenˇ e no. Velk´ e mnoˇ z stv´ ı n´ a hodn´ y ch efekt˚ u m˚ u ˇ z e nˇekdy tice pl´anu pro n´ahodn´e efekty rozmˇer˚ u ni × (l + 1), bi je v´ e st k probl´ e m˚ u m s odhadem modelu a niˇ z ˇ s ´ ı efektivitˇ e odvektor o rozmˇerech (l+1)×1 obsahuj´ıc´ı nezn´ am´e n´ ahodn´e hadovac´ ıho algoritmu, pˇ r edevˇ s ´ ım v pˇ r ´ ıpadˇ e , kdy je soubor efekty, a εi je vektor o rozmˇerech ni ×1 obsahuj´ıc´ı n´ ahodn´e y a/nebo je omezen´ y poˇcet pozorov´an´ı. Pokud se rozsloˇzky. Tedy, pokud jsou prediktory transformace ˇcasu (t) mal´ hodujeme mezi modely s r˚ uzn´ ym poˇctem pevn´ ych efekt˚ u, pomoc´ı FPs, je d˚ uleˇzit´e, aby porovn´avan´e modely mˇely stejn´ y poˇcet n´ahodn´ ych efekt˚ u. V opaˇcn´em pˇr´ıpadˇe nelze jednoznaˇcnˇe yi1 εi1 .. .. rozhodnout, zda volen´ y model je lepˇs´ı z d˚ uvodu zvo . . len´ y ch pevn´ y ch efekt˚ u nebo zvolen´ y ch n´ a hodn´ ych efekt˚ u. yi = (5) Tedy, pokud porovn´av´ame FP1 a FP2 modely, mˇely by yij , εi = εij , . . .. .. oba m´ıt pouze dva n´ahodn´e efekty. Protoˇze v naˇsem pˇr´ıpadˇe oˇcek´av´ame pr´avˇe tento typ srovn´av´an´ı, budeme yini εini d´ale uvaˇzovat modely s pr´avˇe dvˇema n´ahodn´ ymi efekty. β0 p1 pm 1 ti1 . . . ti1 Obecnˇe je FP1 pro LMER d´ an jako .. .. .. , β = β1 , .. Xi = . (6) . . . . . . yij = (β0 + b0i ) + (β1 + b1i )tpij1 + εij . (10) 1 tpin1i . . . tpinmi βm FP2 maj´ı dva exponenty, p1 a p2 , kdy plat´ı, ˇze p1 ≤ p2 . b0i 1 tpi11 . . . tpi1l V pˇr´ıpadˇe, ˇze jsou exponenty rozd´ıln´e, p1 < p2 , m´a FP2 b1i .. .. , b = .. se dvˇema n´ahodn´ ymi efekty formu Zi = ... . (7) . i . . . .. p1 pl 1 tini . . . tini yij = (β0 + b0i ) + (β1 + b1i )tpij1 + β2 tpij2 + εij , (11) bli S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
23
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
zat´ımco v pˇr´ıpadˇe stejn´ ych exponent˚ u, p1 = p2 , je model odhadovan´ ych parametr˚ u v aproximuj´ıc´ım modelu. Proto, d´an jako oˇcek´avanou relativn´ı K-L informaci lze odhadnout pomoc´ı Akaikeho informaˇcn´ıho krit´eria (AIC), kter´e je definov´ano yij = (β0 + b0i ) + (β1 + b1i )tpij1 + β2 tpij2 log tij + εij . (12) jako V tˇechto uveden´ ych FP2 modelech jsou n´ ahodn´e smˇernice d´any vˇzdy prvn´ım typem FP transformace (tj. p1 ). Nicm´enˇe, stejnˇe tak snadno by n´ ahodn´e smˇernice mohly vych´azet z druh´eho typu FP transformace (tj. z p2 ). Kromˇe toho je nutn´e poznamenat, ˇze pokud srovn´av´ame mnoho model˚ u, m˚ uˇze m´ıt p1 rozd´ıln´e hodnoty v FP1 a FP2 modelech, jak uv´ ad´ıme n´ıˇze. V pˇr´ıpadˇe ˇcistˇe explorativn´ı anal´ yzy budeme tedy srovn´avat vˇsechny FP1 a FP2 modely. To znamen´a, ˇze pro uvedenou mnoˇzinu S m´ ame k dispozici devˇet FP1 (9+1)! = 45 FP2 model˚ u. Pro odhad paramodel˚ u a 2!(9−1)! metr˚ u tˇechto 54 model˚ u d´ av´ ame pˇrednost odhadu pomoc´ı ML sp´ıˇse neˇz pomoc´ı REML, protoˇze modely se liˇs´ı pouze v poˇctu pevn´ ych efekt˚ u [18, 28]. Porovn´ av´ an´ı v´ıce model˚ u za pomoci Akaikeho informaˇ cn´ıho krit´ eria V´ ybˇer nejvhodnˇejˇs´ıho modelu (nebo podskupiny model˚ u) mezi celkem 54 odhadnut´ ymi FP1 a FP2 modely je n´aroˇcn´ yu ´kol. Jak jsme jiˇz zm´ınili, v pˇr´ıpadˇe pr˚ uˇrezov´ ych studi´ı a klasick´e line´ arn´ı regrese navrhli Royston a Sauerbrei [4] pouˇz´ıt LRT. V pˇr´ıpadˇe longitudin´ aln´ıch studi´ı s vyuˇzit´ım FPs a LMER je aplikace LRT problematick´a, nebot’ porovn´av´ame navz´ ajem mnoho model˚ u se stejn´ ym poˇctem parametr˚ u a LRT lze pouˇz´ıt pouze pro vnoˇren´e modely. Proto pro v´ ybˇer vhodn´eho modelu z naˇseho souboru 54 LMER model˚ u navrhujeme zcela jinou statistickou strategii zn´amou jako porovn´ av´ an´ı v´ıce model˚ u (multimodel inference) [30]. Tato strategie je rutinnˇe vyuˇz´ıv´ana k porovn´av´an´ı model˚ u v ekologii a biologii volnˇe ˇzij´ıc´ıch zv´ıˇrat [31, 32] a byla ned´ avno prezentov´ ana jako vhodn´a metoda pro anal´ yzu longitudin´ aln´ıch dat v behavior´aln´ıch vˇed´ach [18] a v medic´ınˇe [33, 34]. Porovn´av´an´ı v´ıce model˚ u je statistick´ a koncepce vych´azej´ıc´ı z nov´ ych informaˇcnˇe-teoretick´ ych (I-T) postup˚ u (teorie informace). Tyto I-T metody umoˇzn ˇuj´ı vybrat ten nejvhodnˇejˇs´ı model z a priori zvolen´e mnoˇziny konkuruj´ıc´ıch si model˚ u, tyto modely ohodnotit a seˇradit, co se t´ yˇce vhodnosti, a z´ aroveˇ n odhadnout nejistotu ve v´ ybˇeru modelu [35]. Tyto postupy se naz´ yvaj´ı in” formaˇcnˇe-teoretick´e“, protoˇze jsou zaloˇzeny na Kullbackovˇe-Leiblerovˇe (K-L) teorii informace. K-L informace pˇredstavuje m´ıru informace, kter´ a se ztrat´ı, jestliˇze je dan´ y model pouˇzit´ y jako aproximace skuteˇcn´e reality. Model pˇredstavuj´ıc´ı vlastn´ı realitu je ve skuteˇcnosti nezn´am´ y a proto je naˇs´ım c´ılem vybrat z mnoˇziny moˇzn´ ych model˚ u ten, kter´ y minimalizuje ztr´ atu K-L informace (tj. model, kter´ y je realitˇe nejbliˇzˇs´ı). V roce 1973 Akaike [36] nalezl form´ aln´ı vztah mezi K-L informac´ı a teori´ı vˇerohodnosti. Zjistil, ˇze maximalizovan´ a logaritmick´a vˇerohodnostn´ı funkce je vych´ ylen´ ym odhadem K-L informace a ˇze m´ıra tohoto vych´ ylen´ı je pˇribliˇznˇe rovna poˇctu S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
ˆ AIC = deviance + 2K = −2 log(L(θ|y)) + 2K,
(13)
ˆ kde log(L(θ|y) je hodnota maximalizovan´e logaritmick´e vˇerohodnostn´ı funkce mezi nezn´am´ ymi parametry (θ) pro dan´a data y a K je poˇcet odhadovan´ ych parametr˚ u. AIC penalizuje devianci o dvakr´at poˇcet odhadovan´ ych parametr˚ u, ˇc´ımˇz p˚ usob´ı proti zlepˇsen´ı modelu prost´ ym pˇrid´an´ım bezcenn´eho prediktoru. Pro mal´e v´ ybˇery bychom mˇeli pouˇz´ıvat modifikovan´e AIC (AICcorrected, AICc)([37]), kter´e je v pˇr´ıpadˇe longitudin´aln´ıch dat d´ano 2 · K · (K + 1) , AICc = AIC + P N n − K − 1 i i
(14)
kde suma pˇredstavuje celkov´ y poˇcet vˇsech ˇcasov´ ych bod˚ u pozorov´an´ı. S nar˚ ustaj´ıc´ı velikost´ı v´ ybˇeru se AICc bl´ıˇz´ı AIC. Z tohoto d˚ uvodu by AICc mˇelo b´ yt pouˇz´ıv´ano u mal´ ych i velk´ ych v´ ybˇer˚ u [18]. Na z´akladˇe AICc lze tedy seˇradit modely podle jejich relativn´ı shody modelu s daty a podle jejich hodnovˇernosti. Jestliˇze spoˇc´ıt´ame AICc pro kaˇzd´ y z H model˚ u v naˇs´ı a priori vybran´e mnoˇzinˇe, je potom model s nejniˇzˇs´ım AICc nejl´epe aproximuj´ıc´ım modelem, tj. modelem, kter´ y je nejbl´ıˇze nezn´am´e realitˇe. Hodnota AICc sama o sobˇe interpretovateln´a nen´ı, nebot’ pˇredstavuje relativn´ı shodu modelu s daty. Relativn´ı je proto, ˇze ve skuteˇcnosti nezn´ame prav´ y model a nem˚ uˇzeme tedy ani urˇcit vzd´alenost prav´eho modelu od naˇseho kandid´atn´ıho modelu. Nicm´enˇe, m˚ uˇzeme srovn´avat vz´ajemnˇe si konkuruj´ıc´ı modely mezi sebou a mˇeˇrit, o kolik lepˇs´ı je n´aˇs nejl´epe aproximuj´ıc´ı model ve srovn´an´ı s jin´ ym aproximuj´ıc´ım modelem, tj. m˚ uˇzeme mˇeˇrit velikost efektu. K tomuto u ´ˇcelu bylo navrˇzeno nˇekolik n´astroj˚ u, kter´e n´am maj´ı pomoci pˇri v´ ybˇeru modelu pomoc´ı AIC (a AICc). Nejjednoduˇsˇs´ım zp˚ usobem je zjistit rozd´ıl mezi AICc kaˇzd´eho z konkuruj´ıc´ıch si model˚ u a AICc nejlepˇs´ıho modelu. Necht’ ∆h , pˇredstavuje rozd´ıl pro h-t´ y model. Potom ∆h = AICcmin − AICch ,
(15)
kde h = 1, . . . , H, H je poˇcet model˚ u a AICcmin je nejmenˇs´ı AICc hodnota v mnoˇzinˇe H model˚ u. Nejlepˇs´ı model m´a ∆ = 0 a nejhorˇs´ı model m´a maxim´aln´ı hodnotu ∆. Je moˇzn´e, ˇze pro jakoukoli mnoˇzinu FP model˚ u existuje podmnoˇzina model˚ u s relativnˇe podobnˇe dobrou shodou modelu s daty a podskupina s podobnˇe ˇspatnou shodou. Nen´ı stanoven ˇz´adn´ y jednoznaˇcn´ y cutoff, jak odliˇsit hodnovˇern´e a nehodnovˇern´e podskupiny model˚ u. Ale jako orientaˇcn´ı doporuˇcen´ı n´am m˚ uˇze slouˇzit: modely s ∆h ≤ 2 maj´ı v´ yraznou podporu, modely s 4 ≤ ∆h ≤ 7 maj´ı v´ yraznˇe niˇzˇs´ı podporu a modely s ∆h > 10 nemaj´ı podporu v podstatˇe ˇz´adnou [31].
24
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
Dalˇs´ım n´astrojem pro v´ ybˇer modelu je tzv. v´ aha jistoty (nebo t´eˇz Akaikeho v´ aha) [31]. Tuto v´ ahu jistoty pro h-t´ y model, Wh , lze ch´ apat jako ∆h vyj´ adˇrenou na pravdˇepodobnostn´ı ˇsk´ale, exp(−0.5 · ∆h ) Wh = PH . h exp(−0.5 · ∆h )
(16)
Pokud m´ame data, mnoˇzinu konkuruj´ıc´ıch si model˚ u a nezn´am´ y skuteˇcn´ y model, Wh n´ am potom pˇredstavuje pravdˇepodobnost, ˇze h-t´ y model je nejlepˇs´ım aproximuj´ıc´ım modelem [18]. Model s nejvyˇsˇs´ım W je nejlepˇs´ım aproximuj´ıc´ım modelem, ale m˚ uˇzeme m´ıt v´ıce model˚ u s podobnˇe vysok´ ymi hodnotami. Stejnˇe jako v pˇr´ıpadˇe ∆, ani zde nen´ı d´an jednoznaˇcn´ y cutoff pro hodnovˇernost modelu. Nicm´enˇe, jako rozumn´e hranice byly navrˇzeny Wh = 0, 90 a Wh = 0, 95 [31]. V pˇr´ıpadˇe, ˇze nem´ame ˇz´adn´ y model s takto vysokou pravdˇepodobnost´ı, je vhodn´e vytvoˇrit podmnoˇzinu model˚ u s nejvyˇsˇs´ımi hodnotami W , u nichˇz souˇcet pravdˇepodobnost´ı dosahuje 0,90 nebo 0,95. Tˇret´ım vyj´adˇren´ım velikosti efektu je pomˇ er jistot [31], kter´ y vyjadˇruje, jak mnohem pravdˇepodobnˇejˇs´ı je nejlepˇs´ı model ve srovn´ an´ı s dan´ ym modelem. Pˇredpokl´adejme, ˇze Eh je pomˇer jistot pro h-t´ y model. Potom Wmax , (17) Eh = Wh
ˇcasto setk´av´ame v aplikovan´em medic´ınsk´em v´ yzkumu, obzvl´aˇstˇe v pˇr´ıpadˇe observaˇcn´ıch studi´ı. Nejprve tedy a priori vybereme mnoˇzinu moˇzn´ ych model˚ u a odhadneme jejich parametry. N´aslednˇe s pomoc´ı AICc seˇrad´ıme modely co se t´ yˇce jejich hodnovˇernosti a zhodnot´ıme relativn´ı velikosti efekt˚ u pomoc´ı zm´ınˇen´ ych n´astroj˚ u. Pˇri tomto si je potˇreba uvˇedomit jeden velmi d˚ uleˇzit´ y fakt. V´ ysledek velmi z´avis´ı na tom, kter´e modely vybereme do dan´e mnoˇziny kandid´atn´ıch model˚ u, protoˇze i v pˇr´ıpadˇe, ˇze ˇz´adn´ y z nich nen´ı kvalitn´ı, stejnˇe jsou modely zhodnoceny a seˇrazeny. Vˇzdy bychom proto mˇeli zhodnotit glob´aln´ı hodnotu“ modelu (model˚ u) vy” bran´eho jako nejlepˇs´ı. K tomuto u ´ˇcelu pouˇz´ıv´ame standardn´ı statistick´e n´astroje, tj. indexy absolutn´ı velikosti efektu (napˇr. R2 ) a anal´ yzu rezidu´ı z regresn´ıho modelu. Koncepce porovn´av´an´ı v´ıce model˚ u se stav´ı do protip´olu k tradiˇcn´ı a pouˇz´ıvanˇejˇs´ı metodˇe testov´an´ı nulov´e hypot´ezy (null hypothesis statistical testing, NHST). Vzhledem k tomu, ˇze koncepce porovn´av´an´ı v´ıce model˚ u m˚ uˇze b´ yt pro mnoho ˇcten´aˇr˚ u nov´a, dovolujeme si zd˚ uraznit nˇekter´e z hlavn´ıch odliˇsnost´ı n´astroj˚ u vyuˇz´ıvaj´ıc´ıch AICc ve srovn´an´ı s NHST. • V r´amci paradigmatu I-T metod jsou zaveden´e frekventistick´e principy aplikovateln´e pˇri volbˇe modelu sp´ıˇse neˇz pˇri rozhodov´an´ı o hypot´ez´ach, jak je tomu v pˇr´ıpadˇe NHST. Z tohoto d˚ uvodu nen´ı v prezentovan´e koncepci porovn´av´an´ı v´ıce model˚ u podstatn´e, ˇze prov´ad´ıme mnohon´asobn´a porovn´av´an´ı.
kde Wmax je nejvyˇsˇs´ı v´ aha jistoty v dan´e mnoˇzinˇe model˚ u. Jelikoˇz je Eh pomˇer pravdˇepodobnost´ı, lze ho t´eˇz ch´apat jako ˇsanci, ˇze model h nen´ı nejlepˇs´ım aproximuj´ıc´ım mo• AICc m˚ uˇze porovn´avat nˇekolik model˚ u souˇcasnˇe, delem. Nejlepˇs´ı model m´ a E = 1 a vˇsechny ostatn´ı modely zat´ımco NHST m˚ uˇze vˇzdy hodnotit pouze dva momaj´ı hodnoty vˇetˇs´ı. dely najednou. Pot´e co jsme uvedli tˇri n´ astroje pro porovn´ av´ an´ı mo• AICc lze pouˇz´ıt pro srovn´an´ı vnoˇren´ ych i nedel˚ u zaloˇzen´e na AICc, je nutn´e zm´ınit i nˇekter´ a u ´skal´ı. vnoˇren´ ych model˚ u. Za prv´e, jak jsme jiˇz dˇr´ıve uvedli, vz´ ajemnˇe srovn´avan´e • AICc nepouˇz´ıv´a pro rozhodov´an´ı o modelech armodely by se nemˇely liˇsit v poˇctu n´ ahodn´ ych efekt˚ u bitr´arnˇe stanoven´e hranice, jako je napˇr. p ≤ 0, 05 nebo pˇridruˇzen´ ych komponent variability, jinak n´ am AICc v NHST. Modely lze jednoduˇse seˇradit a rozhom˚ uˇze b´ yt pˇri v´ ybˇeru model˚ u vych´ ylen´ ym vod´ıtkem [38, dov´an´ı je zaloˇzeno na poˇrad´ı bez nutnosti zaho39]. Druh´ ym varov´an´ım je fakt, ˇze aˇckoli AICc bere zen´ı modelu, protoˇze je pod hranic´ı statistick´e v´ yslovnˇe do u ´vahy poˇcet prediktor˚ u, pˇresto m´ a tendenci v´ yznamnosti. preferovat sloˇzitˇejˇs´ı modely [40]. Koncepce vych´ azej´ıc´ı z AICc (a AIC) nen´ı konzistentn´ı v tom smyslu, ˇze zv´ yˇsen´ı • Na z´akladˇe AICc lze jasnˇe vyj´adˇrit velikost efektu, rozsahu v´ ybˇeru nevede ke zv´ yˇsen´ı pravdˇepodobnosti nazat´ımco na z´akladˇe NHST ne. lezen´ı modelu odpov´ıdaj´ı vlastn´ı realitˇe. To samozˇrejmˇe nen´ı probl´em, jestliˇze v´ yzkumn´ık vˇeˇr´ı, ˇze skuteˇcn´ y prav´ y • AICc nevyˇzaduje pˇredpoklad, ˇze alespoˇ n jeden model je nekoneˇcnˇe komplexn´ı a ˇze n´ astroje zaloˇzen´e z kandid´atn´ıch model˚ u je pravdiv´ y, zat´ımco NHST na AICc mu pomohou nal´ezt nejlepˇs´ı aproximuj´ıc´ı moano (v pˇr´ıpadˇe NHST se pˇredpokl´ad´a, ˇze nulov´ y model z mnoˇziny moˇzn´ ych model˚ u. Nicm´enˇe, v pˇr´ıpadˇe ˇze del je pravdiv´ y). v´ yzkumn´ık pˇredpokl´ad´ a, ˇze skuteˇcn´ y model lze odhalit, • N´astroje jako je pomˇer jistot dovoluj´ı urˇcit jistotu, mohou tyto n´astroje v´est k preferenci modelu, kter´ y se ˇze dan´ y model je nejlepˇs´ı, zat´ımco NHST toto nepˇr´ıliˇs pˇrizp˚ usobuje dat˚ um (tzv. overfitting), tj. k volbˇe umoˇ z n ˇ uje. modelu s pˇr´ıliˇs mnoho parametry. T´ım p´ adem filozofick´ y z´amˇer v´ yzkumn´ıka m˚ uˇze ovlivnit stupeˇ n, s jak´ ym budou I-T metody pˇr´ınosn´e. Jestliˇze je c´ılem nal´ezt prav´ y model, Tvorba fet´ aln´ıch percentilov´ ych graf˚ u lze vyuˇz´ıt alternativn´ı indexy (viz [41]). Statistick´a strategie porovn´ av´ an´ı v´ıce model˚ u je Pot´e co jsme prezentovali nˇekolik metod, pokusme pˇrirozen´ ym zp˚ usobem jak pracovat s v´ıce hypot´ezami se shrnout cel´ y postup pro konstrukci fet´aln´ıch refeo moˇzn´ ych modelech z´ aroveˇ n, coˇz je stav, se kter´ ym se renˇcn´ıch kˇrivek na z´akladˇe longitudin´aln´ıch dat. Z´akladem S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
je metoda pr˚ umˇeru a SD, kdy modelujeme pr˚ umˇer a SD fet´aln´ıho biometrick´eho parametru y v z´ avislosti na t (GS) a kdy pro vyj´adˇren´ı vztahu y|t pouˇz´ıv´ ame FP funkce. Na z´akladˇe naˇseho souboru odhadneme mnoˇzinu 54 FP1 a FP2 LMER model˚ u. N´ aslednˇe porovn´ ame vˇsechny modely za pomoci nˇekter´eho z I-T n´ astroj˚ u (napˇr. v´ahy jistoty) a vybereme z mnoˇziny nejlepˇs´ı aproximuj´ıc´ı model. Pot´e co takto z´ısk´ ame fin´ aln´ı model, peˇclivˇe zhodnot´ıme shodu dan´eho modelu s naˇsimi daty. Na z´ avˇer vytvoˇr´ıme na z´akladˇe zvolen´eho modelu vlastn´ı percentilov´e kˇrivky.
SNOMED CT: nenalezen MeSH: nenalezen ICD10: nenalezen
Kullbackova-Leiblerova Informace
Definice: Tento pojem zavedli S. Kullback a R. A. Leibler v roce 1951. Necht’ f pˇredstavuje plnou realitu ˇci skuteˇcnost a g oznaˇcuje aproximuj´ıc´ı model, rozdˇelen´ı pravdˇepodobnosti. Kullbackova-Leiblerova Informace je potom m´ıra ( vzd´alenost“) mezi konZ´ avˇ er ” ceptu´aln´ı realitou, f , a aproximuj´ıc´ım modelem, g, a je definov´ana pro spojit´e rozdˇelen´ı jako integr´al Konstrukce fet´ aln´ıch referenˇcn´ıch graf˚ u vyˇzaduje Z pouˇzit´ı adekv´atn´ı statistick´e metody. V opaˇcn´em pˇr´ıpadˇe f (x) dx I(f, g) = f (x) log mohou b´ yt vytvoˇren´e percentily nepˇresn´e a mohou g(x|θ) v´est k chybn´ ych klinick´ ym z´ avˇer˚ um ohlednˇe v´ yvoje plodu. Pˇredchoz´ı zkuˇsenosti z mnoho prac´ı n´am kde f a g jsou n-dimension´aln´ı rozdˇelen´ı ˇr´ıkaj´ı, ˇze rozdˇelen´ı fet´ aln´ıch biometrick´ ych parametr˚ u je pravdˇepodobnost´ı. K-L informace, oznaˇcen´a I(f, g), pro jak´ekoli GS bl´ızk´e norm´ aln´ımu rozdˇelen´ı. Proto je nejpˇredstavuje m´ıru informace, kterou ztr´ac´ıme, kdyˇz pouˇz´ıvanˇejˇs´ı metodou pro tvorbu fet´ aln´ıch percentilov´ ych model g pouˇzijeme k aproximaci reality, f . Tedy graf˚ u parametrick´ a metoda pr˚ umˇeru a SD s vyuˇzit´ım nejlepˇs´ı model je ten, kter´ y ztr´ac´ı nejm´enˇe inforFPs, kdy modelujeme z´ avislost pr˚ umˇeru a SD dan´eho mace ve srovn´an´ı s ostatn´ımi srovn´avan´ ymi modely, biometrick´eho parametru na GS. Tato pr´ ace prezentuje, neboli ten, kter´ y minimalizuje I(f, g) pro mnoˇzinu jak´ ym zp˚ usobem m˚ uˇze b´ yt st´ avaj´ıc´ı metodika rozˇs´ıˇrena srovn´avan´ ych model˚ u. na longitudin´aln´ı studie pomoc´ı FPs a LMER. Prezentovan´ y postup zahrnuje odhad FP1 a FP2 model˚ u pomoc´ı Synonyma: K-L informace, Kullbackova-Leiblerova divergence, informaˇcn´ı divergence, informaˇcn´ı zisk, reML a v´ ybˇer nejlepˇs´ıho modelu s vyuˇzit´ım strategie polativn´ı entropie rovn´av´an´ı v´ıce model˚ u za pomoci AICc a s n´ım souvisej´ıc´ıch n´astroj˚ u. Zdroj: Burnham KP, Anderson DP. Multimodel inference: Understanding AIC and BIC in model selection. Sociol Methods Res 2004; 33(2): 261–304.
Podˇ ekov´ an´ı
SNOMED CT: nenalezen Tato pr´ace byla podpoˇrena grantem Univerzity KarMeSH: nenalezen lovy v Praze ˇc. SVV-2015-260158. ICD10: nenalezen
Kl´ıˇ cov´ a slova Akaikeho informaˇ cn´ı kriterium Deviance Definice: Deviance je statistika vyjadˇruj´ıc´ı m´ıru shody modelu s daty, v pˇr´ıpadˇe ˇze je model odhadov´an pomoc´ı maxim´ aln´ı vˇerohodnosti. Pˇredstavuje zobecnˇen´ı myˇslenky pouˇzit´ı souˇctu ˇctverc˚ u rezidu´ı v metodˇe nejmenˇs´ıch ˇctverc˚ u na pˇr´ıpady, kdy pr´avˇe odhadujeme parametry modelu pomoc´ı maxim´aln´ı vˇerohodnosti. Deviance je pro dan´ y model na z´akladˇe souboru dat y definov´ ana jako ˆ −2 log(L(θ|y)) ˆ kde log(L(θ|y)) pˇredstavuje hodnotu maximalizovan´e logaritmick´e vˇerohodnostn´ı funkce mezi nezn´am´ ymi parametry (θ) pro dan´ a data y. Zdroj: Nelder JA, Wedderburn RWM. Generalized linear model. J R Statist Soc A 1972; 135: 370–384. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Definice: Akaikeho informaˇcn´ı krit´erium je mˇeˇr´ıtkem relativn´ı kvality statistick´eho modelu pro dan´ y v´ ybˇer a poskytuje prostˇredek k v´ ybˇeru modelu. AIC vych´az´ı z Kullbackovy-Leiblerovy teorie informace: nab´ız´ı relativn´ı odhad ztr´aty informace, kdyˇz pouˇzijeme dan´ y model k vyj´adˇren´ı procesu, kter´ y generuje data. Kromˇe toho lze AIC ch´apat jako nestrann´ y odhad prediktivn´ı pˇrednosti, coˇz je schopnost modelu pˇredpovˇedˇet nov´a data. Toho je dosaˇzeno tak, ˇze AIC balancuje mezi shodou modelu s daty a komplexitou modelu. ˆ AIC = deviance + 2K = −2 log(L(θ|y)) + 2K, kde K je poˇcet parametr˚ u odhadovan´eho modelu a L je maximalizovan´a hodnota vˇerohodnostn´ı funkce pˇr´ısluˇsn´eho modelu. M´ame-li mnoˇzinu kandid´atn´ıch model˚ u pro naˇse data, m´a potom preferovan´ y model
25
26
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
nejniˇzˇs´ı hodnotou AIC. AIC zhodnocuje vhodnost modelu (vyj´adˇren´e pomoc´ı vˇerohodnostn´ı funkce), ale z´aroveˇ n jsou modely penalizov´ any navyˇsov´an´ım AIC v z´avislosti na poˇctu odhadovan´ ych parametr˚ u. Tato penalizace p˚ usob´ı proti pˇr´ıliˇsn´emu se pˇrizp˚ usoben´ı modelu dat˚ um (tzv. overfitting), nebot’ zv´ yˇsen´ı poˇctu parametr˚ u v modelu vˇzdy zlepˇs´ı (sn´ıˇz´ı) devianci. AIC nab´ız´ı metodu, jak lze srovn´avan´e modely zhodnotit a seˇradit a urˇcit relativn´ı velikost efektu modelu, pokud jde o vzd´ alenost modelu od reality.
Inference Definice: Inference je zp˚ usob indukˇcn´ıho uvaˇzov´an´ı, kter´e je zaloˇzeno na informaci z´ıskan´e z v´ ybˇeru, kdy se snaˇz´ıme o zobecnˇen´ı t´eto informace na celou populaci, ze kter´e v´ ybˇer poch´az´ı. Synonyma: indukce, usuzov´an´ı Zdroj: Dodge Y. The Concise Encyclopedia of Statistics. Germany: Springer-Verlag, 2008. SNOMED CT: nenalezen
Synonyma: AIC
MeSH: nenalezen Zdroj: 1. Long J. Longitudinal data analysis for the beICD10: nenalezen havioral sciences using R. Thousand Oaks, Calif.: Sage, 2012. 2. Akaike H. Information theory as an Reference extesion of the maximum likelihood principle. In Petrov BN, Csaki F (Eds). Second international sym- [1] Boulet SL, Salihu HM, Alexander GR. Mode of delivery and birth outcome of macrosomic infants. J Obstet Gynaecol 2004; posium on information theory. Budapest, Hungary: 24: 622–629. Akademiai Kiado, 1973. 3. Burnham KP, Anderson [2] Fang S. Management of preterm infants with intrauterine DR, Huyvaert KP. AIC model selection and mulgrowth restriction. Early Hum Dev 2005; 81: 889–900. timodel inference in behavioral ecology: some bac[3] Hall MH, Carr-Hill RA. The significance of uncertain gestation kground, observations, and comparisons. Behav Ecol for obstetric outcome. Br J Obstet Gynaecol 1985; 92: 452–460. Sociobiol 2011; 65: 23–35. [4] Royston P, Sauerbrei W. Multivariable model-building: A pragSNOMED CT: nenalezen MeSH: nenalezen ICD10: nenalezen
Stupeˇ n volnosti Definice: V kontextu testov´ an´ı pomˇerem vˇerohodnost´ı pˇredstavuje poˇcet stupˇ n˚ u volnosti parametr, kter´ y urˇcuje dan´e referenˇcn´ı χ2 rozdˇelen´ı pouˇzit´e pˇri testov´an´ı hypot´ez. Poˇcet stupˇ n˚ u volnosti je d´an rozd´ılem v poˇctu parametr˚ u, o kter´e se liˇs´ı jednoduˇsˇs´ı redukovan´ y model a komplexnˇejˇs´ı pln´ y model. Tento parametr je pouˇz´ıvan´ y i v dalˇs´ıch pravdˇepodobnostn´ıch rozdˇelen´ıch vztahuj´ıc´ıch se k χ2 rozdˇelen´ı, jako je Studentovo rozdˇelen´ı a Fisherovo rozdˇelen´ı. V jin´e souvislosti se poˇcet stupˇ n˚ u volnosti vztahuje k line´ arnˇe nez´ avisl´ ym ˇclen˚ um pouˇz´ıvan´ ych pˇri v´ ypoˇctu souˇctu ˇctverc˚ u na z´ akladˇe n nez´avisl´ ych pozorov´ an´ı. Pojem stupeˇ n volnosti zavedl R. A. Fisher v roce 1925. Synonyma: df Zdroj: 1. Fisher RA. Applications of Student’s“ dis” tribution. Metron 1925; 5, 90–104. 2. Dodge Y. The Concise Encyclopedia of Statistics. Germany: Springer-Verlag, 2008. SNOMED CT: nenalezen
matic approach to regression analysis based on fractional polynomials for modelling continuous variables. Chichester, England: John Wiley, 2008. [5] Wright EM, Royston P. A comparison of statistical methods for age-related reference intervals. J R Statist Soc A 1997; 160: 47–69. [6] Silverwood RJ, Cole TJ. Statistical methods for constructing gestational age-related reference intervals and centile charts for fetal size. Ultrasound Obstet Gynecol 2007; 29: 6–13. [7] Hynek M. Approaches for constructing age-related reference intervals and centile charts for fetal size. Eur J Biomed Informatics 2010; 6: 43–52. [8] Borghi E, Onis M, Garza C, et. al. Construction of the World Health Organization child growth standards: selection of methods for attained growth curves. Statist Med 2006; 25: 247– 265. [9] Altman DG, Chitty LS. Design and analysis of studies to derive charts of fetal size. Ultrasound Obstet Gynecol 1993; 3: 378–384. [10] Villar J, Altman DG, Purwar M, et al.; International Fetal and Newborn Growth Consortium for the 21st Century. The objectives, design and implementation of the INTERGROWTH21st Project. BJOG 2013; 120 (Suppl 2): 9–26. [11] Ioannou C, Talbot K, Ohuma E, et al. Systematic review of methodology used in ultrasound studies aimed at creating charts of fetal size. BJOG 2012; 119(12): 1425–39. [12] Altman DG, Ohuma EO; International Fetal and Newborn Growth Consortium for the 21st Century. Statistical considerations for the development of prescriptive fetal and newborn growth standards in the INTERGROWTH-21st Project. BJOG 2013; 120 (Suppl 2): 71–76. [13] Royston P, Altman DG. Regression using fractional polynomials of continuous covariates: Parsimonious parametric modelling. Appl Statist 1994; 43: 429–467.
MeSH: nenalezen
[14] Altman DG. Construction of age-related reference centiles using absolute residuals. Statistics in Medicine 1993; 12: 917– 924.
ICD10: nenalezen
[15] Royston P, Wright EM. How to construct ’normal ranges’ for fetal variables. Ultrasound Obstet Gynecol 1998; 11: 30–38.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Hynek M. a kol. – Statistick´e metody pro tvorbu ˇcasovˇe z´avisl´ych percentilov´ych graf˚ u ...
[16] Aitkin MA. Modelling variance heterogenity in normal regression using GLIM. Applied Statistics 1987; 36: 332–339.
[29] Laird N, Ware J. Random-effects models for longitudinal data. Biometrics 1982; 38: 963–974.
[17] Mosteller F, Tukey J. Data analysis and regression: A second course. New York: Addison-Wesley, 1977.
[30] Burnham KP, Anderson DR. Model Selection and Multimodel Inference. New York: Springer, 2002.
[18] Long JD. Longitudinal data analysis for the behavioral sciences using R. Thousand Oaks, Calif.: Sage, 2012.
[31] Burnham KP, Anderson DR. Multimodel inference: understanding AIC and BIC in model selection. Socio Meth Res 2004; 33(2): 261–304.
[19] Chitty LS, Altman DG. Charts of fetal size: limb bones. BJOG 2002; 109: 919–929. [20] Papageorghiou AT, Kennedy SH, Salomon LJ, et al.; International Fetal and Newborn Growth Consortium for the 21st Century (INTERGROWTH-21st). International standards for early fetal size and pregnancy dating based on ultrasound measurement of crown-rump length in the first trimester of pregnancy. Ultrasound Obstet Gynecol 2014; 44: 641–648. [21] Papageorghiou AT, Ohuma EO, Altman DG, et al.; International Fetal and Newborn Growth Consortium for the 21st Century (INTERGROWTH-21st). International standards for fetal growth based on serial ultrasound measurements: the Fetal Growth Longitudinal Study of the INTERGROWTH-21st Project. Lancet 2014; 384(9946): 869–879. [22] Royston P, Wright EM. Goodness-of-fit statistics for agespecific reference intervals. Stat Med 2000; 19: 2943–2962. [23] D’Agostino RB, Belanger A, D’Agostino Jr RB. A Suggestion for Using Powerful and Informative Tests of Normality. Am Stat 1990; 44: 316–321. [24] Sasieni P, Royston P, Cox NJ. Symmetric nearest neighbour linear smoothers. Stata J 2005; 5: 285. [25] Friedman JH, Silverman BW. Flexible Parsimonious Smoothing and Additive Modeling. Technometrics 1989; 31(1): 3–21. [26] Long J, Ryoo Jihoon. Using fractional polynomials to model non-linear trends in longitudinal data. Br J Math Stat Psychol 2010; 63: 177–203.
[32] Burnham KP, Anderson DR, Huyvaert KP. AIC model selection and multimodel inference in behavioral ecology: some background, observations, and comparisons. Behav Ecol Sociobiol 2011; 65: 23–35. [33] Paulsen JS, Smith MM, Long JD; PREDICT HD investigators and Coordinators of the Huntington Study Group. Cognitive decline in prodromal Huntington: Disease: implications for clinical trials. J Neurol Neurosurg Psychiatry 2013; 84: 1233–1239. [34] Leoni V, Long JD, Mills JA, et al.; PREDICT-HD study group. 24S-hydroxycholesterol correlation with markers of Huntington disease progression. Neurobiol Dis 2013; 55: 37–43. [35] Burnham KP, Anderson DR. Kullback-Leibler information as a basis for strong inference in ecological studies. Wildl Res 2001; 28: 111–119. [36] Akaike H. Information theory as an extension of the maximum likelihood principle. In: Petrov BN, Csaki F, editors. Proceedings of the 2nd International Symposium on Information Theory; 1973; Budapest, Hungary: Akademiai Kiado. p. 267–281. [37] Hurvich CM, Tsai CL. Regression and time series model selection in small samples. Biometrika 1989; 76: 297–307. [38] Liang H, Wu H, Zou G. A note on conditional AIC for linear mixed-effects models. Biometrica 2008; 95: 773–778. [39] Vaida F, Blanchard S. Conditional Akaike information for mixed-effects model. Biometrica 2005; 92: 351–370.
[27] Fitzmaurice GM, Laird NM, Ware JH. Applied longitudinal analysis. New York: Wiley, 2004.
[40] Link WA, Barker RJ. Model weights and the foundations of multimodel inference. Ecology 2006; 87: 2626–2635.
[28] Galecki A, Burzykowski T. Linear mixed-effects models using R a step-by-step approach. New York: Springer, 2013.
[41] Kadane JB, Lazar NA. Methods and criteria for model selection. J Amer Statist Assoc 2004; 99: 279–290.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
27
28
ˇarek M. – V´yznam deficitu cerebr´aln´ıho fol´atu pro v´yvoj a progresi autismu Krsiˇcka D., S´
V´ yznam deficitu cerebr´ aln´ıho fol´ atu pro v´ yvoj a progresi autismu ˇ arek2 Daniel Krsiˇ cka1 , Milan S´ 1
ˇ a republika 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova, Praha, Cesk´ 2
ˇ a republika EuroMISE Mentor Association, Praha, Cesk´
Kontakt: Daniel Krsiˇ cka 1. l´ ekaˇrsk´ a fakulta, Univerzita Karlova, Praha ˇ Adresa: Kateˇrinsk´ a 32, 121 08, Praha, CR E–mail:
[email protected]
C´ıle v´ yzkumu Naˇs´ım c´ılem je lepˇs´ı porozumˇen´ı vztah˚ um mezi autismem, resp. poruchou autistick´eho spektra (PAS), a deficitem cerebr´aln´ıho fol´atu (CFD), jeho etiologie, patofyziologie a synergick´ ych efekt˚ u nedostatku fol´ at˚ u v r˚ uzn´ ych kritick´ ych okamˇzic´ıch v´ yvoje jedince a dalˇs´ıch subklinick´ ych nox z prostˇred´ı nebo genetick´ ych predispozic. Nˇekter´e geny asociovan´e s PAS nebo jin´ ymi neurov´ yvojov´ ymi poruchami maj´ı metabolickou roli, protoˇze jsou pˇrekl´ad´any do proteinu s katalytickou aktivitou – enzymu. Pˇredpokl´ad´ame, ˇze nedostatek fol´ at˚ u m˚ uˇze negativnˇe ovlivnit nˇekter´e metabolick´e reakce kritick´e pro rozvoj nebo progresi PAS podobn´ ym zp˚ usobem jako patologick´e zmˇeny v dan´ ych genech dokonce i za podm´ınek, kdy jsou dan´e geny intaktn´ı, nebo synergisticky zesilovat v´ yznam m´enˇe v´ yznamn´ ych genetick´ ych patologi´ı. Tedy, ˇze dysregulace v mnoˇzstv´ı substr´ at˚ u, metabolit˚ u nebo kofaktor˚ u mohou rezultovat v podobnou poruchu jako dysfunkˇcn´ı polymorfismy nebo delece pˇr´ısluˇsn´ ych gen˚ u za norm´aln´ıch koncentrac´ı tˇechto l´ atek. Tak´e by mohlo b´ yt zaj´ımav´e zkoumat dalˇs´ı patologick´e efekty nedostatku fol´at˚ u, jako jsou zmˇeny v genov´e expresi na podkladˇe dysregulace koncentrace fol´ at˚ u, zmˇeny ve v´ yvojov´e signalizaci, zmˇeny v methylaˇcn´ıch vzorech DNA nebo vztahy k ˇc´asteˇcn´e mitochondri´ aln´ı dysfunkci. Naˇse hypot´eza se zakl´ad´a na publikovan´e klinick´e zkuˇsenosti s diagn´ozou a l´eˇcbou konkomitantn´ıho CFD u PAS s hl´ aˇsen´ ym zm´ırnˇen´ım nebo dokonce potlaˇcen´ım jadern´ ych symptom˚ u PAS v nˇekter´ ych pˇr´ıpadech. N´aˇs prvn´ı c´ıl je deterministicky nal´ezt vztahy mezi fol´atov´ ym metabolismem a metabolick´ ymi reakcemi asociovan´ ymi s autismem v co nejˇsirˇs´ım moˇzn´em z´abˇeru a urˇcit mezi r˚ uzn´ ymi publikovan´ ymi biochemick´ ymi, fyziologick´ ymi a klinick´ ymi n´ alezy ty, kter´e jsou skuteˇcnˇe v´ yznamn´e pro rozvoj a progresi PAS. Takov´a znalost, kdyˇz bude vytvoˇrena, by mohla b´ yt d´ale rozv´ıjena v hlubˇs´ım detailu napˇr´ıklad stran v´ yznamu deplece fol´at˚ u pro pˇresnˇe definovan´e prenat´ aln´ı nebo postnat´aln´ı v´ yvojov´e f´aze nebo v´ yznamu nedostatku fol´at˚ u k vyˇsˇs´ı prevalenci PAS u chlapc˚ u a mnoha dalˇs´ıch. Tak´e
m˚ uˇzeme testovat, jak se zmˇen´ı exprese gen˚ u za podm´ınek sn´ıˇzen´e koncentrace fol´ at˚ u, tj. kter´e geny budou v´ıce exprimov´any a zdali nˇekter´e z nich, na z´akladˇe aktu´alnˇe dostupn´ ych informac´ı, nejsou asociov´any s PAS. Seznam moˇzn´ ych experiment˚ u je mnohem ˇsirˇs´ı a je tak´e obohacen pˇrekryvem s dalˇs´ımi onemocnˇen´ımi jako je schizofrenie, amn´ezie nebo trisomie 21, kter´e se jev´ı tak´e ˇc´asteˇcnˇe z´avisl´e na fol´atech.
Souˇ casn´ y stav pozn´ an´ı Etiologie autismu, resp. poruch autistick´eho spektra (PAS), je v´ıcem´enˇe nezn´am´a, genetick´e syndromy pˇripadaj´ı pouze na zhruba 15% pˇr´ıpad˚ u. Etiologie je pˇredevˇs´ım multifaktori´aln´ı, a z toho d˚ uvodu jak´ ykoli efektivn´ı v´ yzkum mus´ı zkoumat v´ıce faktor˚ u. V naˇsem posledn´ım review [1] jsme naˇsli celkem 351 publikovan´ ych pˇr´ıpad˚ u CFD s r˚ uznˇe z´avaˇzn´ ym PAS u 44% pacient˚ u a 56% z tˇechto pˇr´ıpad˚ u bylo zp˚ usobeno protil´atkami fol´atov´ ych receptor˚ u (FRAA), kter´e naruˇsuj´ı, nebo dokonce blokuj´ı pˇrenos fol´at˚ u pˇres hematoencefalickou bari´eru. Speci´alnˇe se soustˇred´ıme na tyto pˇr´ıpady PAS, protoˇze bˇeˇznˇe je pomˇernˇe obt´ıˇzn´e a nepravdˇepodobn´e diagnostikovat m´ırn´ y CFD u pacient˚ u s PAS. Jedin´a dostupn´a a zcela spolehliv´a metoda je lumb´aln´ı punkce s vyˇsetˇren´ım hladiny 5-MTHF (5-methyltetrahydrofol´at) v likvoru, kter´a nen´ı, pro svou invazivitu, rutinnˇe indikov´ana u PAS. M´enˇe invazivn´ı metody, jako napˇr. MR spektroskopie, dnes nemaj´ı potˇrebnou rozliˇsovac´ı schopnost v ˇr´adech nmol/L. Vyˇsetˇren´ı na FRAA pozitivitu je zat´ım dostupn´e jen v nˇekolika laboratoˇr´ıch na svˇetˇe. Proto je moˇzn´e, ˇze m´ırn´ y CFD u PAS dlouhodobˇe unik´a pozornosti v diagnostice a l´eˇcbˇe a ˇze m˚ uˇze pˇrisp´ıvat k rozvoji a progresi PAS. Nav´ıc, 29% nalezen´ ych pˇr´ıpad˚ u PAS s konkomitantn´ım CFD ukazuje, ˇze l´eˇcba m˚ uˇze zm´ırnit nebo obˇcas dokonce potlaˇcovat jadern´e symptomy PAS [2, 3, 4, 5, 6, 7]. Nemˇelo by b´ yt opomenuto, ˇze existuj´ı tak´e jin´e experiment´aln´ı studie l´eˇcby PAS, zaloˇzen´e na pod´av´an´ı l´atek, jejichˇz synt´eza nebo koncentrace je pˇr´ımo ˇci nepˇr´ımo z´avisl´a na S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
ˇarek M. – V´yznam deficitu cerebr´aln´ıho fol´atu pro v´yvoj a progresi autismu Krsiˇcka D., S´
fol´atech nebo fol´ aty kontrolovan´ ych reakc´ıch, ve kter´ ych fol´aty vystupuj´ı jako kofaktory [8, 9, 10, 11, 12, 13]. V´ yznamnˇe zv´ yˇsen´e hladiny FRAA rezultuj´ı v typick´ y degenerativn´ı progreduj´ıc´ı obraz CFD, nicm´enˇe v´ yznam m´ırnˇe zv´ yˇsen´ ych hladin FRAA, resp. m´ırnˇe sn´ıˇzen´ ych hladin fol´at˚ u pouze v CNS, zat´ım nen´ı pro patofyziologii PAS dostateˇcnˇe zn´am. Abnormality v hladin´ ach fol´at˚ u mohou pravdˇepodobnˇe pˇrisp´ıvat k mnoha dalˇs´ım epigenetick´ ym naruˇsen´ım spojovan´ ym s PAS a dalˇs´ımi neurov´ yvojov´ ymi a neurodegenerativn´ımi poruchami, jako je abnorm´aln´ı methylace genomu, oxidativn´ı stress, mitochondri´aln´ı a neuron´aln´ı poˇskozen´ı nebo abnorm´ aln´ı imunitn´ı odpovˇed’. Samotn´ y dlouhodob´ y nedostatek fol´ at˚ u zˇrejmˇe nen´ı potˇreba, ale p˚ usob´ıc´ı spoleˇcnˇe s dalˇs´ımi faktory jako nev´ yznamn´e genetick´e mutace, subklinick´a environment´aln´ı toxick´a z´ atˇeˇz, dlouhodob´ y stres nebo specifick´e protil´atky jako tˇreba FRAA mohou vysvˇetlit a zp˚ usobovat abnorm´aln´ı r˚ ust CNS, neuron´ aln´ı diferenciaci, migraci a pruningu nebo aktivaci apopt´ ozy vedouc´ı k neurodegeneraci. Ve vztahu k neuron´ aln´ı tk´ ani nen´ı dlouhodob´ y inzult potˇreba. Vliv kr´ atko- nebo stˇrednˇedob´eho nedostatku fol´at˚ u v r˚ uzn´ ych v´ yvojov´ ych obdob´ıch jedince nebyly zat´ım dostateˇcnˇe studov´ any.
hou manifestovat jin´ ym zp˚ usobem a nemus´ı b´ yt prim´arnˇe spojov´any s PAS v klinick´e praxi. Fol´atovˇe-dependentn´ı fenotyp PAS, pokud bude identifikov´an, by mohl v´ yznamnˇe ovlivnit dalˇs´ı klinick´ y v´ yzkum l´eˇcby PAS podle pravidel medic´ıny zaloˇzen´e na d˚ ukazech. Pro efektivn´ı rozliˇsen´ı fenotyp˚ u, screening, vˇcasnou diagnostiku a biologick´e terapie PAS je nezbytn´e l´epe pochopit synergick´ y efekt p˚ usoben´ı v´ıce faktor˚ u, kter´e spoleˇcnˇe manifestuj´ı jako neurobehavior´aln´ı syndrom, ale samostatnˇe kaˇzd´ y zvl´aˇst’ pˇredstavuj´ı jen subklinick´ y nemanifestn´ı probl´em. Heterogenn´ı pˇr´ıˇciny, kter´e mohou p˚ usobit souˇcasnˇe a synergicky posilovat jedna druhou, pokud nejsou zkoum´any spoleˇcnˇe, prakticky znemoˇzn ˇuj´ı spr´avnˇe vybrat testovac´ı skupinu do jak´ekoli studie. Sm´ıˇsen´a testovac´ı skupina sloˇzen´a z v´ıce vz´ajemnˇe nerozliˇsen´ ych fenotyp˚ u PAS, by ovlivnila v´ ysledek kaˇzd´e studie n´ahodn´ ym zp˚ usobem a mezi v´ ysledky studi´ı by byly vˇzdy v´ yrazn´e rozd´ıly. Postupnˇe by bylo moˇzn´e identifikovat pouze dominantn´ı pˇr´ıˇciny, jak se tomu zd´a b´ yt u z´avaˇzn´eho CFD na podkladˇe FRAA. Naopak, pokud by bylo moˇzn´e identifikovat a klasifikovat jednotliv´e, samostatnˇe nev´ yznamn´e, ale spoleˇcnˇe manifestn´ı noxy, v´ yraznˇe by to pˇrispˇelo k identifikaci nov´ ych etiologi´ı PAS. Proto Prevalence m´ırn´eho CFD u PAS nen´ı dostateˇcnˇe zma- identifikace a klasifikace kaˇzd´eho nov´eho fenotypu PAS je yvoj nezbytn´a. pov´ana stejnˇe jako jeho ovlivnˇen´ı dalˇs´ıch patologick´ ych pro dalˇs´ı v´ pˇr´ıˇcin. Byla publikov´ ana jen velmi mal´ a ˇc´ ast pˇr´ıpad˚ u (celkem 351). Dokonce i vliv terapie m´ırn´eho CFD na Uplatnˇ en´ı v biomedic´ınˇ e symptomy PAS nen´ı dostateˇcnˇe dokumentov´ an. Publikoı van´e v´ ysledky jasnˇe dokazuj´ı potˇrebu dalˇs´ıho v´ yzkumu a zdravotnictv´ v t´eto oblasti pˇredevˇs´ım proto, ˇze PAS je aktu´alnˇe povaˇzov´an za nel´eˇciteln´e celoˇzivotn´ı postiˇzen´ı nezn´am´e M´ame pˇredbˇeˇzn´e v´ ysledky, kter´e potvrzuj´ı naˇse hyetiologie s vysok´ ym socioekonomick´ ym zat´ıˇzen´ım. U pa- pot´ezy. Pro tyto v´ ysledky stran vztah˚ u fol´at˚ u a PAS exiscient˚ u s PAS a prok´azan´ ym CFD byly opakovanˇe po- tuj´ı 2 hlavn´ı praktick´a vyuˇzit´ı: pisov´ana zlepˇsen´ı neurologick´ ych symptom˚ u i jadern´ ych symptom˚ u PAS. Nˇekter´e publikace tak´e ukazuj´ı negativn´ı • prevence korelaci mezi vˇekem pacienta a u ´ˇcinnost´ı l´eˇcby. Ran´a intervence se tedy jev´ı jako zcela z´ asadn´ı. • l´eˇcba Aktu´aln´ı v´ yzkum pˇr´ıˇcin PAS sest´ av´ a z 2 hlavn´ıch proud˚ u. Prvn´ı zkoum´ a polygenn´ı vliv mnoha gen˚ u a konU PAS jako celoˇzivotn´ıho, obecnˇe nel´eˇciteln´eho pocentruje se na genom jako takov´ y. Druh´ y smˇer v´ yzkumu stiˇzen´ı je rozhoduj´ıc´ı identifikovat suspektn´ı tˇehotenstv´ı se zab´ yv´a epigenetick´ ymi zmˇenami a p˚ usoben´ım fak- nebo dokonce rodiˇce co nejdˇr´ıve. Jestliˇze existuje vztah tor˚ u prostˇred´ı. Oba tak pˇredstavuj´ı v´ yzvu z oblasti mezi nedostatkem fol´at˚ u a nˇekter´ ymi pomˇernˇe nov´ ymi noBig Data jako zpracov´ an´ı velkoobjemov´ ych dat z perso- xami jako FRAA ovlivˇ nuj´ıc´ımi vyv´ıjej´ıc´ı se plod nebo d´ıtˇe, nalizovan´ ych celogenomov´ ych sekvenov´ an´ı, porovn´av´an´ı mus´ı b´ yt identifikov´an, aby byla moˇzn´a u ´ˇcinn´a prevence velk´ ych mnoˇzin dat, automatizace odvozov´ an´ı nov´ ych in- nebo kompenzace probl´emu pˇredt´ım, neˇz dojde k rozvoji formac´ı a dalˇs´ı. PAS. S ohledem na nˇekolik posledn´ıch v´ ysledk˚ u se zd´a, ˇze existuje fol´atovˇe-dependentn´ı fenotyp PAS a ˇze nedostatek fol´at˚ u tak´e m˚ uˇze pˇrisp´ıvat k v´ yvoji nebo zhorˇsen´ı dalˇs´ıch onemocnˇen´ı jako schizofrenie [14, 15, 16], n´ahl´a amn´ezie v dospˇelosti [17] nebo trisomie 21 [18]. Tyto fenotypy mohou sest´ avat z polygenn´ıho vlivu v´ıce faktor˚ u, kter´e jsou synergisticky posilov´ any nedostatkem fol´at˚ u. Genetick´a sloˇzka se zd´ a b´ yt st´ ale d˚ uleˇzit´a, nicm´enˇe nem˚ uˇze pˇresvˇedˇcivˇe vysvˇetlit vˇsechny pˇr´ıpady a v´ ysledky v´ yzkumu. Zd´a se, ˇze faktory prostˇred´ı formuj´ıc´ı epigenetick´e vlivy by mˇely b´ yt vzaty v u ´vahu. Tyto faktory moS´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Pˇredpokl´ad´ame, ˇze lepˇs´ı a hlubˇs´ı pochopen´ı vztah˚ u mezi fol´aty a patofyziologi´ı PAS m˚ uˇze odkr´ yt dalˇs´ı genetick´e, epigenetick´e a environment´aln´ı rizikov´e faktory, kter´e mohou pˇrisp´ıvat k v´ yznamu m´ırn´e deplece fol´at˚ u, tedy faktory definuj´ıc´ı fol´atovˇe-fragiln´ı“ fenotyp PAS. ” Tyto informace by mohly b´ yt pouˇzity k v´ yvoji nov´ ych populaˇcn´ıch screening˚ u a celkovˇe v´est ke sn´ıˇzen´ı prevalence PAS a jeho socioekonomick´eho zat´ıˇzen´ı. Neposledn´ım vyuˇzit´ım v´ ysledk˚ u v´ yzkumu je st´ale otevˇren´a moˇznosti vyuˇzit´ı fol´at˚ u jako souˇc´asti biologick´e terapie PAS u jiˇz rozvinut´ ych pˇr´ıpad˚ u.
29
30
ˇarek M. – V´yznam deficitu cerebr´aln´ıho fol´atu pro v´yvoj a progresi autismu Krsiˇcka D., S´
Podˇ ekov´ an´ı
Fol´ atov´ y receptor 1
y je Tato pr´ace byla podpoˇrena projektem SVV-2015- Definice: Subtyp GPI-pˇripojen´eho receptoru, kter´ exprimov´an v tk´an´ıch epiteli´aln´ıho p˚ uvodu, kli260158 Univerzity Karlovy v Praze. nicky v´ yznamn´ y v oblasti plexus choroideus kv˚ uli 3n´asobn´e koncentraci fol´at˚ u v cerebrospin´aln´ım proKl´ıˇ cov´ a slova storu ve srovn´an´ı s hladinou v krvi.
Autismus
Synonyma: FOLR1 receptor
Zdroj: https://en.wikipedia.org/wiki/Folate_ Definice: Pervazivn´ı neurov´ yvojov´ a porucha charakterireceptor_1 zovan´a postiˇzen´ım komunikaˇcn´ıch dovednost´ı, dysfunkˇcn´ı soci´aln´ı interakc´ı, deficitem pˇredstavivosti SNOMED CT: nenalezeno a stereotypn´ım chov´ an´ım. MeSH: D12.776.157.530.450.074.500.299.500.500 Synonyma: Porucha autistick´eho spektra, dˇetsk´ y autisICD10: nenalezeno mus (nepˇresn´e) Zdroj: https://en.wikipedia.org/wiki/Autism
Protil´ atka fol´ atov´ ych receptor˚ u
SNOMED CT: 408856003
Definice: Protil´atka typu IgM nebo IgG proti proteinu FOLR1 naruˇsuj´ıc´ı nebo blokuj´ıc´ı vysokoafinitn´ı n´ızkokoncentraˇcn´ı receptor FOLR1 v´ yznamnˇe exprimovan´ y na hematoencefalick´e bari´eˇre.
MeSH: F03.550.325.125 ICD10: nenalezen pˇresn´ y k´ od (pouze nepˇresn´ y F84.0)
Syndrom deficitu cerebr´ aln´ıho fol´ atu
Synonyma: FOLR1 protil´atka
Zdroj: http://www.ncbi.nlm.nih.gov/pubmed/ 23314538 Definice: Progresivn´ı neurodegenerativn´ı syndrom, pˇredevˇs´ım v ran´em dˇetstv´ı, charakterizovan´ y SNOMED CT: nenalezeno sn´ıˇzenou hladinou fol´ at˚ u v centr´ aln´ım nervov´em syst´emu pˇri souˇcasnˇe norm´ aln´ıch syst´emov´ ych hla- MeSH: nenalezeno din´ach. ICD10: nenalezeno Synonyma: Neurodegenerace na podkladˇe nedostatku cerebr´aln´ıch fol´at˚ u Reference Zdroj: http://www.ncbi.nlm.nih.gov/pubmed/ 15581159 SNOMED CT: nenalezeno MeSH: nenalezeno ICD10: pˇresn´ y k´ od nenalezen (pouze nepˇresn´ y E53.9)
Fol´ at Definice: Jedna z biologicky aktivn´ıch forem kyseliny listov´e vz´ajemnˇe se odliˇsuj´ıc´ıch r˚ uznou u ´rovn´ı oxidace/redukce a poˇctem glutamov´ ych zbytk˚ u. Synonyma: Kyselina listov´ a, Kyselina Pteroylglutamov´a, Vitamin B9 Zdroj: https://en.wikipedia.org/wiki/Folic_acid SNOMED CT: 63718003 MeSH: D03.438.733.631.400 ICD10: nenalezeno
[1] D. Krsiˇ cka, M. Vlˇ ckov´ a, and M. Havlovicov´ a, “The Significance of Cerebral Folate Deficiency for the Development and Treatment of Autism Spectrum Disorders,” Int. J. Biomed. Healthc., vol. 1, no. 1, 2015. [2] V. T. Ramaekers, S. P. Rothenberg, J. M. Sequeira, T. Opladen, N. Blau, E. V Quadros, and J. Selhub, “Autoantibodies to folate receptors in the cerebral folate deficiency syndrome.,” N. Engl. J. Med., vol. 352, no. 19, pp. 1985–1991, May 2005. [3] V. T. Ramaekers, J. M. Sequeira, N. Blau, and E. V Quadros, “A milk-free diet downregulates folate receptor autoimmunity in cerebral folate deficiency syndrome.,” Dev. Med. Child Neurol., vol. 50, no. 5, pp. 346–52, May 2008. [4] P. Moretti, S. U. Peters, D. Del Gaudio, T. Sahoo, K. Hyland, T. Bottiglieri, R. J. Hopkin, E. Peach, S. H. Min, D. Goldman, B. Roa, C. a Bacino, and F. Scaglia, “Brief report: autistic symptoms, developmental regression, mental retardation, epilepsy, and dyskinesias in CNS folate deficiency.,” J. Autism Dev. Disord., vol. 38, no. 6, pp. 1170–1177, Jul. 2008. [5] S. U. Steele, S. M. Cheah, A. Veerapandiyan, W. Gallentine, E. C. Smith, and M. A. Mikati, “Electroencephalographic and seizure manifestations in two patients with folate receptor autoimmune antibody-mediated primary cerebral folate deficiency.,” Epilepsy Behav., vol. 24, no. 4, pp. 507–12, Aug. 2012. [6] R. E. Frye, J. M. Sequeira, E. V Quadros, S. J. James, and D. A. Rossignol, “Cerebral folate receptor autoantibodies in autism spectrum disorder.,” Mol. Psychiatry, vol. 18, no. 3, pp. 369–81, Mar. 2013.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
ˇarek M. – V´yznam deficitu cerebr´aln´ıho fol´atu pro v´yvoj a progresi autismu Krsiˇcka D., S´
[7] R. S. Al-Baradie and M. W. Chaudhary, “Diagnosis and management of cerebral folate deficiency. A form of folinic acidresponsive seizures.,” Neurosciences (Riyadh)., vol. 19, no. 4, pp. 312–6, Oct. 2014.
[13] K. Williams, A. Brignell, M. Randall, N. Silove, and P. Hazell, “Selective serotonin reuptake inhibitors (SSRIs) for autism spectrum disorders (ASD).,” Cochrane database Syst. Rev., vol. 8, p. CD004677, Jan. 2013.
[8] D. A. Rossignol and R. E. Frye, “Melatonin in autism spectrum disorders: a systematic review and meta-analysis.,” Dev. Med. Child Neurol., vol. 53, no. 9, pp. 783–92, Sep. 2011.
[14] V. T. Ramaekers, B. Th¨ ony, J. M. Sequeira, M. Ansseau, P. Philippe, F. Boemer, V. Bours, and E. V Quadros, “Folinic acid treatment for schizophrenia associated with folate receptor autoantibodies.,” Mol. Genet. Metab., Oct. 2014.
[9] R. Coben, M. Linden, and T. E. Myers, “Neurofeedback for autistic spectrum disorder: a review of the literature.,” Appl. Psychophysiol. Biofeedback, vol. 35, no. 1, pp. 83–105, 2010. [10] K. Bertoglio, S. Jill James, L. Deprey, N. Brule, and R. L. Hendren, “Pilot study of the effect of methyl B12 treatment on behavioral and biomarker measures in children with autism.,” J. Altern. Complement. Med., vol. 16, no. 5, pp. 555–560, May 2010. [11] R. E. Frye, D. Rossignol, M. F. Casanova, G. L. Brown, V. Martin, S. Edelson, R. Coben, J. Lewine, J. C. Slattery, C. Lau, P. Hardy, S. H. Fatemi, T. D. Folsom, D. Macfabe, and J. B. Adams, “A Review of Traditional and Novel Treatments for Seizures in Autism Spectrum Disorder: Findings from a Systematic Review and Expert Panel.,” Front. public Heal., vol. 1, p. 31, 2013. [12] E. A. Langley, M. Krykbaeva, J. K. Blusztajn, and T. J. Mellott, “High maternal choline consumption during pregnancy and nursing alleviates deficits in social interaction and improves anxiety-like behaviors in the BTBR T+Itpr3tf/J mouse model of autism.,” Behav. Brain Res., Oct. 2014.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
[15] A. Ho, D. Michelson, G. Aaen, and S. Ashwal, “Cerebral folate deficiency presenting as adolescent catatonic schizophrenia: a case report.,” J. Child Neurol., vol. 25, no. 7, pp. 898–900, 2010. [16] Q. Wang, J. Liu, Y.-P. Liu, X.-Y. Li, Y.-Y. Ma, T.-F. Wu, Y. Ding, J.-Q. Song, Y.-J. Wang, and Y.-L. Yang, “[Methylenetetrahydrofolate reductase deficiency-induced schizophrenia in a school-age boy].,” Zhongguo Dang Dai Er Ke Za Zhi, vol. 16, no. 1, pp. 62–6, Jan. 2014. [17] Z. Sadighi, I. J. Butler, and M. K. Koenig, “Adult-onset cerebral folate deficiency.,” Arch. Neurol., vol. 69, no. 6, pp. 778–779, 2012. [18] H. Blehaut, C. Mircher, A. Ravel, M. Conte, V. de Portzamparc, G. Poret, F. H. de Kermadec, M.-O. Rethore, and F. G. Sturtz, “Effect of leucovorin (folinic acid) on the developmental quotient of children with Down’s syndrome (trisomy 21) and influence of thyroid status.,” PLoS One, vol. 5, no. 1, p. e8394, Jan. 2010.
31
32
Muˇzn´y M. et al. – Integrace zaˇr´ızen´ı pro l´eˇcbu diabetu a lifestyle zaˇr´ızen´ı v r´amci mobiln´ı aplikace . . .
Integrace zaˇr´ızen´ı pro l´ eˇ cbu diabetu a lifestyle zaˇr´ızen´ı v r´ amci mobiln´ı aplikace pro self-management diabetu Miroslav Muˇ zn´ y1,2 , Martina Vlas´ akov´ a1 , Jan Muˇ z´ık1,2 , Eirik Arsand3 1 2 3
ˇ a republika 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova v Praze, Cesk´
ˇ ˇ a republika Fakulta biomedic´ınsk´eho inˇzen´yrstv´ı, CVUT v Praze, Cesk´ Norsk´e centrum pro integrovanou p´eˇci a telemedic´ınu, Tromso, Norsko
Kontakt: Miroslav Muˇ zn´ y Centrum podpory aplikaˇ cn´ıch v´ ystup˚ u a spin-off firem, 1. l´ ekaˇrsk´ a fakulta, Univerzita Karlova v Praze Adresa: Studniˇ ckova 7, 128 08 Praha 2 E–mail:
[email protected]
C´ıle v´ yzkumu Lid´e trp´ıc´ı onemocnˇen´ım diabetes mellitus maj´ı moˇznost l´epe zvl´adat jeho pr˚ ubˇeh pouˇz´ıv´ an´ım r˚ uzn´ ych zaˇr´ızen´ı, kter´a jim pom´ ahaj´ı l´epe porozumˇet souvislostem, v jak´ ych reaguje jejich tˇelo. Toto je znaˇcnˇe usnadnˇeno dostupnost´ı modern´ıch chytr´ ych telefon˚ u a digit´ aln´ıch diabetick´ ych den´ık˚ u, kter´e funguj´ı jako agreg´ ator a interpret sesb´ıran´ ych dat. D˚ uleˇzitou roli v tomto adaptaˇcn´ım procesu hraj´ı nositeln´ a zaˇr´ızen´ı, kter´ a integruj´ı nˇekolik typ˚ u senzor˚ u a mohou tak efektivnˇe spojit funkcionalitu nˇekolika zaˇr´ızen´ı. Integrace z nˇekolika datov´ ych zdroj˚ u nicm´enˇe pˇrin´aˇs´ı urˇcit´e riziko zneuˇzit´ı dat. Zapracov´ an´ı dat do digit´aln´ıho den´ıku diabetika mus´ı b´ yt proto zajiˇstˇeno tak, aby nehrozila kompromitace funkc´ı dan´eho zaˇr´ızen´ı a jejich pouˇz´ıv´an´ı neohrozilo samotn´eho uˇzivatele [1]. Jak´ y typ dat m˚ uˇze b´ yt zaj´ımav´ y pro uˇzivatele digit´aln´ıho den´ıku diabetika? Pˇr´ıkladem jsou data o fyzick´e aktivitˇe, nutriˇcn´ı data, data o tepov´e frekvenci, v´aze a data z inzul´ınov´ ych pump a kontinu´ aln´ıch glukometr˚ u. Tato data mohou b´ yt sb´ıran´ a pomoc´ı sn´ımaˇc˚ u fyzick´e aktivity, chytr´ ych hodinek, vah a pomoc´ı closed-loop syst´em˚ u. Kaˇzd´e ze zaˇr´ızen´ı, kter´e je zaintegrov´ ano do digit´aln´ıho den´ıku diabetika, m˚ uˇze poskytnout nov´ y pohled na self-management diabetu pacienta. Vzhledem k absenci bezpeˇcn´eho, otevˇren´eho komunikaˇcn´ıho rozhran´ı je nutn´e peˇclivˇe zv´aˇzit rizika plynouc´ı pro jejich uˇzivatele. V r´amci naˇseho v´ yzkumu zkoum´ ame moˇznosti integrace podobn´ ych zaˇr´ızen´ı do mobiln´ı aplikace Diabetesdagboka [2].
kontrolu nad vlastn´ımi daty. V posledn´ı dobˇe v´ yrobci tˇechto zaˇr´ızen´ı umoˇzn ˇuj´ı nahr´avat nasb´ıran´a data na vzd´alen´ y server, kter´ y slouˇz´ı pro potˇreby z´alohy a z´aroveˇ n jako prostˇredek pro vzd´alenou synchronizaci s v´ıce zaˇr´ızen´ımi. Maj´ı vˇsak uˇzivatel´e z´ajem o to, aby jejich zdravotn´ı data byla na vzd´alen´ ych serverech? V´ yrobci lifestyle zaˇr´ızen´ı a zaˇr´ızen´ı pro l´eˇcbu diabetu obecnˇe neposkytuj´ı prostˇredky kter´e by umoˇznily jejich bezpeˇcnou integraci do aplikac´ı tˇret´ıch stran pomoc´ı pˇr´ım´e komunikace s dan´ ym zaˇr´ızen´ım a omezuj´ı tak jejich uˇzivatele k pouˇzit´ı propriet´arn´ıho software. Aˇckoliv jsou nˇekter´a zaˇr´ızen´ı pro tento u ´ˇcel uzp˚ usobena, pro vˇetˇsinu z nich je pˇr´ım´e komunikaˇcn´ı rozhran´ı nedostupn´e.
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı V posledn´ıch mˇes´ıc´ıch lze pozorovat d˚ uleˇzitost komunit, kter´e se zab´ yvaj´ı vytv´aˇren´ım alternativn´ıho softwaru pro zaˇr´ızen´ı na trhu. Jedn´ım z nich je komunita Nightscout, jej´ıˇz ˇclenov´e se na internetu identifikuj´ı pomoc´ı tagu #WeAreNotWaiting. Lid´e v r´amci t´eto komunity vyvinuli mobiln´ı aplikaci pro kontinu´aln´ı glukometry Dexcom, kterou je moˇzn´e vyˇc´ıtat data pomoc´ı bˇeˇznˇe dostupn´eho vybaven´ı (USB OTG kabelu). Kontinu´aln´ı glukometr je pak moˇzn´e vyuˇz´ıt napˇr´ıklad pro vzd´alen´e sledov´an´ı stavu glyk´emie d´ıtˇete.
Dalˇs´ım pˇr´ıkladem je Diabeto, startup projekt kter´ y si klade za c´ıl vyvinout zaˇr´ızen´ı slouˇz´ıc´ı pro jednoduˇsˇs´ı integraci glukometr˚ u v r´amci mobiln´ı aplikace. Toto zaˇr´ızen´ı bude moˇzn´e pˇripojit do jack konektoru glukometru a poSouˇ casn´ y stav pozn´ an´ı moc´ı nˇej bezdr´atovˇe vyˇc´ıtat namˇeˇren´e glykemick´e hodnoty do chytr´eho telefonu. Podobn´ y pˇr´ıstup by znaˇcnˇe uzn´ ych glukometr˚ u a jejich Jedn´ım z nejv´ıce diskutovan´ ych omezen´ı pro uˇzivatele usnadnil pouˇzit´ı nˇekolika r˚ vyuˇz´ıvaj´ıc´ı zaˇr´ızen´ı pro l´eˇcbu diabetu, je nemoˇznost m´ıt bezpeˇcnou integraci do mobiln´ıho diabetick´eho den´ıku. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
33
Muˇzn´y M. et al. – Integrace zaˇr´ızen´ı pro l´eˇcbu diabetu a lifestyle zaˇr´ızen´ı v r´amci mobiln´ı aplikace . . .
D´ıky v´ yˇse popsan´ ym snah´ am komunitn´ıho v´ yvoje v uzavˇren´e smyˇcce, kter´ y d´avkuje inzul´ın autoje moˇzn´e podstatnˇe snadnˇeji integrovat zaˇr´ızen´ı do dimaticky na z´akladˇe dat ze senzoru pro sledov´an´ı git´aln´ıho diabetick´eho den´ıku. Pˇrestoˇze tyto modifikace glyk´emie. a u ´pravy nejsou ofici´ alnˇe podporov´ any v´ yrobci zaˇr´ızen´ı y infuzn´ı syst´em pro l´eˇcbu diabetu a jejich pouˇzit´ı vyˇzaduje technick´e Synonyma: Inzul´ınov´ znalosti, ukazuj´ı nov´e moˇznosti, jak vyuˇz´ıt tato zaˇr´ızen´ı Zdroj: National Library of Medicine v r˚ uzn´ ych sc´en´aˇr´ıch bˇeˇzn´eho ˇzivota pro lepˇs´ı l´eˇcbu diaSNOMED CT: 69805005 betu.
Podˇ ekov´ an´ı
MeSH: D007332 ICD10: nenalezeno
Tato pr´ace byla podpoˇrena projektem SVV-2015Sn´ımaˇ c fyzick´ e aktivity 260158 Univerzity Karlovy v Praze.
Kl´ıˇ cov´ a slova
Definice: Nositeln´ y senzor slouˇz´ıc´ı k odhadu fyzick´e aktivity.
Nositeln´ a technika
Synonyma: Senzor fyzick´e aktivity
ˇ asti obleˇcen´ı a pˇr´ısluˇsenstv´ı, kter´e integruje Definice: C´ prostˇredky v´ ypoˇcetn´ı techniky.
Zdroj: Yang, Che-Chang, and Yeh-Liang Hsu. A re” view of accelerometry-based wearable motion detectors for physical activity monitoring.“ Sensors 10.8 (2010): 7772-7788.
Synonyma: Wearables, Tech togs, Nositeln´e zaˇr´ızen´ı Zdroj: https://en.wikipedia.org/wiki/Wearable_ technology
SNOMED CT: nenalezeno MeSH: nenalezeno
SNOMED CT: nenalezeno MeSH: nenalezeno
ICD10: nenalezeno
ICD10: nenalezeno
Vzd´ alen´ y dohled pacient˚ u
Closed-loop syst´ em
Definice: Komunikaˇcn´ı s´ıt’ pro poskytov´an´ı u ´daj˚ u o zdravotn´ım stavu pacienta ze vzd´alen´eho m´ısta.
Definice: Syst´em pro automatick´e d´ avkov´ an´ı inzul´ınu, Synonyma: Vzd´alen´ y dohled kter´ y je sloˇzen ze tˇr´ı komponent: podkoˇzn´ıho senzoru pro sledov´ an´ı glyk´emie, algoritmu pro v´ ypoˇcet Zdroj: Rosenfeld, Brian, and Michael Breslow. Tele” d´avkov´an´ı inzul´ınu a intraperitonealn´ı inzul´ınov´e communications network for remote patient moniinfuzn´ı pumpy. toring.“ U.S. Patent No. 7,256,708. 14 Aug. 2007. Synonyma: Umˇel´ a slinivka
SNOMED CT: nenalezeno
Zdroj: Renard, Eric, et al. Closed-loop insulin delivery MeSH: nenalezeno ” using a subcutaneous glucose sensor and intraperitoneal insulin delivery feasibility study testing a new ICD10: nenalezeno model for the artificial pancreas.“ Diabetes Care 33.1 (2010): 121-127. Reference SNOMED CT: 261000004 MeSH: nenalezeno ICD10: nenalezeno
Inzul´ınov´ a pumpa Definice: Syst´em pro d´ avkov´ an´ı inzul´ınu. Zahrnuje syst´em pro regulaci v otevˇren´e smyˇcce, kter´ y m˚ uˇze b´ yt kontrolov´ an pacientem ˇci pˇreddefinovan´ ym programem a je navrˇzen pro neust´ al´e d´ avkov´an´ı mal´ ych d´avek inzul´ınu, kter´e jsou zv´ yˇsen´e v pr˚ ubˇehu tr´aven´ı. Z´aroveˇ n zahrnuje syst´em pro regulaci S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
[1] KLONOFF, David C. Cybersecurity for Connected Diabetes Devices. Journal of diabetes science and technology, 2015, 1932296815583334. [2] ARSAND, Eirik, SKROVSETH, Stein Olav, JOAKIMSEN, Ragnar Martin, HARTVIGSEN, Gunnar. Design of an Advanced Mobile Diabetes Diary Based on a Prospective 6-month Study Involving People with Type 1 Diabetes. The 6th International Conference on Advanced Technologies and Treatments for Diabetes, February 27. - March 2. 2013, Paris. France. [3] Misfit Scientific Library. https://build.misfit.com/
Available
from:
[4] The Nightscout Project. http://www.nightscout.info
Available
from:
[5] Diabeto. Available from: http://diabe.to/
34
Rak D., Sv´atek V. – Porovn´av´an´ı WS a l´ekaˇrsk´ych DP s vyuˇzit´ım datab´az´ı ˇr´ızen´ych medic´ınsk´ych slovn´ık˚ u
Porovn´ av´ an´ı internetov´ ych str´ anek a l´ ekaˇrsk´ ych doporuˇ cen´ ych postup˚ u s vyuˇ zit´ım datab´ az´ı ˇr´ızen´ ych medic´ınsk´ ych slovn´ık˚ u Duˇsan Rak1 , Vojtˇ ech Sv´ atek2 1 2
ˇ a republika 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova v Praze, Praha, Cesk´
ˇ a republika Fakulta informatiky a statistiky, Vysok´ a ˇskola ekonomick´ a v Praze, Praha, Cesk´
Kontakt: Duˇsan Rak Adresa: U Z´ atiˇs´ı 545/9, 14700, Praha 4 E–mail:
[email protected]
C´ıle v´ yzkumu
Doporuˇcen´e postupy (DP) v l´ekaˇrstv´ı jsou systematicky vytv´aˇreny a publikov´any renomovan´ ymi l´ekaˇrsk´ ymi spoleˇcnostmi na z´akladˇe v´ ysledk˚ u medic´ıny zaloˇzen´e na d˚ ukazech (MZD) [2]. Tyto texty jsou n´aslednˇe publikov´any veˇrejnˇe v strukturovan´e textov´e podobˇe urˇcen´e pro vyuˇzit´ı odbornou veˇrejnost´ı. DP jsou nav´ıc tak´e dostupn´e ve vysoce formalizovan´e podobˇe – napˇr´ıklad ve form´atu GLIF [3]. C´ılem t´eto formalizace je usnadnit n´asledn´e vyuˇzit´ı informac´ı obsaˇzen´ ych v DP automatizovan´ ymi informatick´ ymi postupy.
Prob´ıhaj´ıc´ı v´ yzkum m´ a za c´ıl pˇredloˇzit a otestovat jednoduch´ y postup hodnocen´ı kvality obsahu webov´ ych str´anek (WS) s vyuˇzit´ım doporuˇcen´ ych postup˚ u (DP) ˇ ızen´e medic´ınsk´e slovn´ıky zde jako standardu kvality. R´ slouˇz´ı jako prostˇredek, pomoc´ı kter´eho jsou obˇe skupiny dokument˚ u porovn´av´any. Z obou skupin porovn´ avan´ ych dokument˚ u (WS a DP) jsou extrahov´ any mnoˇziny pouˇzit´ ych odborn´ ych term´ın˚ u, kter´e jsou dohled´ av´any pr´avˇe v ˇr´ızen´ ych medic´ınsk´ ych slovn´ıc´ıch, jako je UMLS nebo MeSH. Kvalita WS je stanovov´ ana jednak na z´ akladˇe obecnˇejˇs´ı shody obsahu (tedy vyskytuj´ıc´ıch se koncept˚ u en´ı v biomedic´ınˇ e a t´emat) a d´ale na z´akladˇe podobnosti konkr´etn´ı pouˇzit´e Uplatnˇ odborn´e terminologie s koncepty a terminologi´ı vysky- a zdravotnictv´ ı tuj´ıc´ı se v DP. D´ılˇc´ım c´ılem je navrhnout a vyhodnotit vhodn´e metody agregace terminologie DP tak, aby bylo moˇzno testovan´e WS porovn´ avat v˚ uˇci jedin´emu mˇeˇr´ıtku DP pro jednotliv´e oblasti medic´ıny poskytuj´ı cenn´ y i v pˇr´ıpadˇe, ˇze je dostupn´ ych DP v´ıce. D˚ uleˇzit´ ym c´ılem standard kvality vyuˇziteln´ y odbornou lekaˇrskou veˇrejnost´ı pr´ace je vyhodnocen´ı pouˇzitelnosti tohoto postupu v pro- a tak´e certifikaˇcn´ımi autoritami pro ruˇcn´ı hodnocen´ı kvacesu poloautomatick´eho hodnocen´ı kvality. lity obsahu webov´ ych str´anek. I pˇres vysokou formalizaci DP je ale jejich vyuˇzit´ı pro pln´e, zcela automatizovan´e hodnocen´ı kvality obsahu zat´ım nemysliteln´e. Souˇ casn´ y stav pozn´ an´ı Pˇredpokl´adalo by totiˇz velmi pokroˇcil´e strojov´e poroModern´ı technologie poskytuj´ı ˇradu moˇznost´ı jak pu- zumˇen´ı textu. blikovat na internetu komukoli t´emˇeˇr cokoli. To se t´ yk´a i medic´ınsk´ ych informac´ı, kde se ˇsirok´ a dostupnost mnoha Na druhou stranu n´ami testovan´e relativnˇe pˇr´ımoˇcar´e informac´ı s takto z´avaˇzn´ ym obsahem ve velmi variabiln´ı porovn´an´ı podobnosti mnoˇzin nalezen´ ych odborn´ ych kvalitˇe, st´av´a velk´ ym probl´emem. Specializovan´e organi- term´ın˚ u (do urˇcit´e m´ıry jistˇe reprezentuj´ıc´ıch obsah), mezi zace zab´ yvaj´ıc´ı se hodnocen´ım kvality webov´ ych str´anek DP a hodnocen´ ymi WS, m˚ uˇze b´ yt realisticky automatis medic´ınskou problematikou a jejich certifikac´ı (napˇr. zovateln´ ym, ale souˇcasnˇe velmi cenn´ ym a vypov´ıdaj´ıc´ım HON [1]), kter´e dˇr´ıve hodnotily webov´e str´ anky ruˇcn´ı re- mˇeˇr´ıtkem kvality. S jeho pomoc´ı lze z´ıskat pˇredstavu cenz´ı, vyv´ıjej´ı n´astroje na automatizaci nebo alespoˇ n po- napˇr´ıklad o u ´plnosti textu, kvalitˇe a odbornosti pouˇzit´e loautomatizaci vyhodnocov´ an´ı krit´eri´ı kvality. Automati- terminologie, pˇr´ıpadnˇe WS klasifikovat do skupin pozace je ovˇsem v pˇrev´aˇzn´e vˇetˇsinˇe zat´ım omezena pouze na dobnosti. Vedle tˇechto kvantitativn´ıch krit´eri´ı je moˇzno hodnocen´ı form´aln´ıch aspekt˚ u kvality prezentac´ı a jedin´a lidsk´ ym expert˚ um, zab´ yvaj´ıc´ım se hodnocen´ım a certifire´aln´a moˇznost, jak ovˇeˇrit kvalitu obsahu medic´ınsk´ ych kac´ı WS, poskytnout jako pˇridanou hodnotu takt´eˇz vlastn´ı text˚ u, z˚ ust´av´a verifikace odborn´ıkem v oboru. texty WS anotovan´e odbornou terminologi´ı. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Rak D., Sv´atek V. – Porovn´av´an´ı WS a l´ekaˇrsk´ych DP s vyuˇzit´ım datab´az´ı ˇr´ızen´ych medic´ınsk´ych slovn´ık˚ u
Podˇ ekov´ an´ı
Zdroj: Sackett DL. Evidence based medicine: what it is and what it isn’t. BMJ 1996;312:7
Tato pr´ace byla podpoˇrena projektem SVV-2015SNOMED CT: nenalezeno 260158 Univerzity Karlovy v Praze.
Kl´ıˇ cov´ a slova
MeSH: D019317 ICD10: nenalezeno
Anotace Podobnost Definice: Anotac´ı rozum´ıme opatˇren´ı zdrojov´ ych dat ˇuj´ıc´ı pˇriˇradit skupinˇe doku(textu, videa, obrazu atd.) popisn´ ymi nebo klasi- Definice: Koncept umoˇzn ment˚ u nebo skupinˇe seznam˚ u term´ın˚ u hodnotu ukafikaˇcn´ımi z´aznamy. Napˇr´ıklad jde o pˇrid´ an´ı r˚ uzn´ ych zuj´ıc´ı vz´ajemnou bl´ızkost ˇci shodu jejich s´emantiky. pozn´amek, znaˇcek nebo obecnˇe jak´ ychkoli metainformac´ı k dokumentu jako celku nebo pˇr´ımo k jeho Synonyma: S´emantick´a / obsahov´a podobnost ˇci ˇc´astem. bl´ızkost Synonyma: Lingvistick´ a anotace (ve zpracov´an´ı textu), Zdroj: pˇreloˇzeno ze str´anky http://en.wikipedia. Tagov´an´ı org/wiki/Semantic_similarity Zdroj: kombinovan´ a definice z v´ıce zdroj˚ u SNOMED CT: nenalezeno SNOMED CT: nenalezeno MeSH: nenalezeno
MeSH: nenalezeno
ICD10: nenalezeno
ICD10: nenalezeno
Koncept
UMLS
mnoha ˇr´ızen´ ych biomeDefinice: Koncept je z´ akladn´ı jednotkou v´ yznamu – Definice: Kompendium dic´ınsk´ ych slovn´ık˚ u. P˚ uvodnˇe bylo vytvoˇreno v roce nˇekdy je povaˇzov´ an za z´ akladn´ı sloˇzku znalost´ı. 1986 v NLM (americkou N´arodn´ı l´ekaˇrskou kniKaˇzd´ y koncept se skl´ ad´ a z dalˇs´ıch ˇc´ ast´ı, kter´e jsou ´ celem UMLS je usnadnˇen´ı v´ hovnou). Uˇ yvoje jeho vlastnostmi. Koncept tedy nen´ı jen definic´ı, ale v´ ypoˇcetn´ıch syst´em˚ u, kter´e se chovaj´ı jako by rosp´ıˇse oznaˇcen´ım pro skupinu vlastnost´ı nˇeˇceho, co ” zumˇely“ v´ yznamu pˇrirozen´eho jazyka v biomedic´ınˇe pˇredstavuje. a zdravotnictv´ı. S t´ım c´ılem NLM vytv´aˇr´ı a disSynonyma: Pojem (v kognitivn´ı vˇedˇe) tribuuje UMLS Knowledge sources“ (znalostn´ı ” zdroje ve formˇe datab´az´ı) a asociovan´e softwarov´e Zdroj: pˇreloˇzeno a upraveno z http://en. n´astroje pro vyuˇzit´ı v´ yvoj´aˇri. Existuj´ı tˇri UMLS wikipedia.org/wiki/Concept a http://www. ” Knowledge Sources“: the Metathesaurus“ (metatethefreedictionary.com/concept ” zaurus), the Semantic Network“ (s´emantick´a s´ıt’) ” SNOMED CT: nenalezeno a SPECIALIST Lexicon“. ” MeSH: nenalezeno Zdroj: NML On-Line ICD10: nenalezeno
Medic´ına zaloˇ zen´ a na d˚ ukazech (MZD)
SNOMED CT: nenalezeno MeSH: D017432
Definice: MZD vˇedomˇe, c´ılenˇe a explicitnˇe vyuˇz´ıv´a ICD10: nenalezeno aktu´aln´ı nejlepˇs´ı a nejsilnˇejˇs´ı vˇedeck´e d˚ ukazy pro rozhodov´an´ı o zdravotn´ı p´eˇci pro konkr´etn´ı pacienty. Reference Praxe MZD integruje expertn´ı znalosti l´ekaˇre s nejlepˇs´ımi dostupn´ ymi d˚ ukazy v´ yzkumu. Zohledˇ nov´ana [1] Health On the Net Foundation [Internet]. Available from: http://www.hon.ch/ je s´ıla d˚ ukaz˚ u o rizic´ıch a pˇr´ınosech jednotliv´ ych l´eˇcebn´ ych postup˚ u (vˇcetnˇe nel´eˇcen´ı) o diagnos- [2] Sackett DL. Evidence based medicine: what it is and what it isn’t. BMJ 1996;312:7 tick´ ych testech atd. Synonyma: Praxe zaloˇzen´ a na d˚ ukazech (v´ yznamovˇe ˇsirˇs´ı pojem)
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
[3] Ohno-Machado L, et al. The GuideLine Interchange Format: A Model for Representing Guidelines. J Am Med Inform Assoc. 1998 Jul-Aug; 5(4): 357–372.
35
36
Seidl L., Hanzl´ıˇcek P. – Mezin´arodn´ı komunikaˇcn´ı standardy a interoperabilita syst´em˚ u v ˇcesk´em zdravotnictv´ı
Mezin´ arodn´ı komunikaˇ cn´ı standardy a interoperabilita syst´ em˚ u vˇ cesk´ em zdravotnictv´ı Libor Seidl1 , Petr Hanzl´ıˇ cek2 1
ˇ a republika 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova v Praze, Praha, Cesk´ 2
ˇ Cargo, Praha, Cesk´ ˇ a republika CD
Kontakt: Libor Seidl Centrum pro eHealth a telemedic´ınu Adresa: Studniˇ ckova 7, Praha 2, 120 00 E–mail:
[email protected]
C´ıle v´ yzkumu C´ılem v´ yzkumu je identifikace moˇznost´ı a posouzen´ı pˇrek´aˇzek pˇri aplikaci mezin´ arodn´ıch komunikaˇcn´ıch protokol˚ u do ˇcesk´eho prostˇred´ı. ˇ e zdravotnictv´ı se na poli interoperability inCesk´ formaˇcn´ıch syst´em˚ u op´ır´ a zejm´ena o ˇcesk´ y datov´ y standard DASTA [1]. Tento protokol vznik´ a od roku 1997 dle aktu´aln´ıch potˇreb jednotliv´ ych v´ yrobc˚ u nemocniˇcn´ıch informaˇcn´ıch syst´em˚ u na realizaci pˇrenosu dat. Pˇri zbˇeˇzn´em pohledu na specifikaci DASTA zcela chyb´ı definice spouˇstˇec´ıch ud´ alost´ı (kdy se datov´ y obsah komunikuje), v´ yznam pˇren´ aˇsen´ ych dat (nov´ y z´aznam/smaz´an´ı z´ aznamu), definice zodpovˇednost´ı aplikaˇcn´ıch rol´ı odes´ılatele a pˇr´ıjemce a nen´ı uvedena ani n´asledn´a kask´ada dalˇs´ıch interakc´ı. Sp´ıˇse neˇz o zpr´avovˇe orientovan´ y komunikaˇcn´ı protokol se tedy jedn´ a o standardizovanou datovou strukturu pro zdravotnictv´ı, kter´a se pravdˇepodobnˇe pouˇz´ıv´ a obdobnˇe jako dokument. Mezin´arodn´ı standardy naproti tomu nab´ızej´ı na zpr´av´ach zaloˇzen´e komunikaˇcn´ı protokoly (HL7 verze 2, HL7 verze 3) i definice dokument˚ u (Clinical Document Architecture – CDA, Continuity of Care Record – CCR a EN 13606). Vzhledem k rozˇs´ıˇrenosti implementac´ı protokolu DASTA v ˇcesk´em zdravotnictv´ı lze oˇcek´ avat, ˇze pˇri postupn´em uplatnˇen´ı mezin´ arodn´ıch standard˚ u bude nutn´e alespoˇ n po pˇrechodnou dobu prov´ adˇet transformaci datov´ ych tok˚ u mezi vˇsemi datov´ ymi protokoly. V tomto procesu zmˇeny se budou pr˚ ubˇeˇznˇe mˇenit komunikaˇcn´ı sc´en´aˇre, coˇz vyˇzaduje vysokou flexibilitu v konfiguraci meziprotokolov´e br´any. Tato br´ ana tedy nem˚ uˇze b´ yt realizov´ana jako relativnˇe jednoduch´ a jedno´ uˇcelov´ a transformace, nebot’ neust´al´e zmˇeny a testov´ an´ı implementace by byly n´akladn´e a ˇcasov´a n´ aroˇcnost upgrade by mohla brzdit celkov´ y zmˇenov´ y proces ve zdravotnictv´ı. V´ ychodiskem se jev´ı automaticky konfigurovan´a br´ana, kter´a vyuˇz´ıv´a formalizovan´e znalosti jak o inte-
graˇcn´ım prostˇred´ı, o protokolech, tak o komunikovan´e klinick´e dom´enˇe. Ontologick´e popisy datov´ ych komunikaˇcn´ıch protokol˚ u a jejich vz´ajemn´e mapov´an´ı umoˇzn´ı objektivn´ı posouzen´ı protokolu DASTA k dalˇs´ım mezin´arodn´ım standard˚ um. D´ale bude moˇzn´e posoudit praktiˇcnost pouˇzit´ı a identifikovat pˇrek´aˇzky pˇri aplikaci mezin´arodn´ıch komunikaˇcn´ıch protokol˚ u do ˇcesk´eho prostˇred´ı.
Souˇ casn´ y stav pozn´ an´ı Zp˚ usob popisu architektury syst´emu popisuje norma ISO/IEC/IEEE 42010 z roku 2011, kter´a nahradila p˚ uvodn´ı normu IEEE 1471:2000. V t´eto normˇe je tak´e uveden konceptu´aln´ı model, kter´ y zasazuje syst´em do jeho okol´ı – kaˇzd´ y syst´em existuje v nˇejak´em okol´ı a naplˇ nuje nˇejak´ yu ´ˇcel, ze kter´eho profituj´ı uˇzivatel´e (stakeholdeˇri). Generick´ y komponentn´ı model (GCM)[2] popisuje architekturu obecn´eho informaˇcn´ıho syst´emu rozdˇelen´ım popisu na dom´enovou perspektivu, perspektivu architektury syst´emu a perspektivu v´ yvoje a implementace software. Oddˇelen´e popisy soused´ıc´ıch ˇc´ast´ı GCM krychle jsou n´aslednˇe provazov´any transformaˇcn´ımi mechanizmy. Ontologickou reprezentaci komunikaˇcn´ıch protokol˚ u HL7 verze 2 a HL7 verze 3 poprv´e navrhl Frank Oemig ve sv´e disertaˇcn´ı pr´aci [3]. Pro anal´ yzu a popis d´ılˇc´ıch ˇc´ast´ı probl´emu pouˇzil pr´avˇe architekturn´ı pˇr´ıstup podle GCM. Pro popis struktury vlastn´ıch protokol˚ u pouˇzil Frank Oemig vyˇsˇs´ı ontologii CSO (Communication Standards Ontology)[4] a automaticky generovan´e definice zpr´av v OWL z datab´azov´eho popisu HL7 v2 a z MIF soubor˚ u popisuj´ıc´ıch zpr´avy HL7 v3. Pro popis dom´eny (administrativn´ı popis pacienta) byla pouˇzita referenˇcn´ı ontologie ACGT (Advanced Clinico-Genomic Trial Ontology). Celkov´e sch´ema sestaven´ı ontologi´ı je zobrazeno na obr´azku 1. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Seidl L., Hanzl´ıˇcek P. – Mezin´arodn´ı komunikaˇcn´ı standardy a interoperabilita syst´em˚ u v ˇcesk´em zdravotnictv´ı
Obr´ azek 1: Sestaven´ı ontologi´ı, pˇrevzato z [3].
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı
Kl´ıˇ cov´ a slova Komunikaˇ cn´ı protokol
Definice: Aplikaˇcn´ı protokol pro elektronickou v´ ymˇenu V´ yzkum by mˇel identifikovat vhodn´e existuj´ıc´ı ontodat v prostˇred´ı zdravotnictv´ı. logie, doplnit je, pˇr´ıpadnˇe vytvoˇrit doplˇ nuj´ıc´ı ontologie, aby vznikl ucelen´ y popis integraˇcn´ıho prostˇred´ı, klinick´e Synonyma: Komunikaˇcn´ı standard dom´eny i komunikaˇcn´ıch protokol˚ u. Zdroj: http://skmtglossary.org/search.aspx? term_id=1161&SearchExp=communication% V´ yzkum by tak´e mˇel uk´ azat cestu, jak odliˇsn´e ontolo20protocol gie jednotliv´ ych ˇc´ ast´ı GCM propojovat v jedin´ y funkˇcn´ı popis re´aln´eho prostˇred´ı. Tento ucelen´ y popis bude vyuˇzit SNOMED CT: nenalezeno pro automatickou konfiguraci gateway i zhodnocen´ı zralosti protokol˚ u, vhodnosti pro dan´e integraˇcn´ı prostˇred´ı MeSH: nenalezeno a identifikaci pˇrek´ aˇzek pro pouˇzit´ı mezin´ arodn´ıch stanICD10: nenalezeno dard˚ u v ˇcesk´em prostˇred´ı.
DASTA
Aplikovan´ ym v´ ystupem v´ yzkumu by mohla b´ yt i technologick´a demonstrace automaticky konfigurovaDefinice: DASTA je pravidelnˇe aktualizovan´ y, otevˇren´ y teln´e mezi-protokolov´e br´ any pro vybran´ y pˇr´ıpad. Komstandard pro komunikaci mezi informaˇcn´ımi binace formalizovan´e znalosti a interpretu t´eto formalisyst´emy zdravotnick´ ych zaˇr´ızen´ı, kter´ y pokr´ yv´a obzace m˚ uˇze uk´azat budouc´ı zp˚ usob tvorby komplexn´ıch inlasti klinick´e, laboratorn´ı, statistick´e i administraformaˇcn´ıch syst´em˚ u. tivn´ı a jehoˇz samozˇrejmou souˇc´ast´ı jsou ˇc´ıseln´ıky (napˇr´ıklad N´arodn´ı ˇc´ıseln´ık laboratorn´ıch poloˇzek, ´ ˇc´ıseln´ık klinick´ ych ud´alost´ı, aktu´aln´ı ˇc´ıseln´ıky UZIS, atd.), dokumenty a n´ a stroje (napˇ r ´ ıklad program Podˇ ekov´ an´ı ˇ CLP). Synonyma: Datov´ y standard Tato pr´ace byla podpoˇrena projektem SVV-2015260158 Univerzity Karlovy v Praze. Zdroj: http://www.dastacr.cz/ S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
37
38
Seidl L., Hanzl´ıˇcek P. – Mezin´arodn´ı komunikaˇcn´ı standardy a interoperabilita syst´em˚ u v ˇcesk´em zdravotnictv´ı
SNOMED CT: nenalezeno
Br´ ana
MeSH: nenalezeno
Definice: Entita, kter´a v re´aln´em ˇcase zabezpeˇcuje obousmˇernou komunikaci mezi koncov´ ymi stanicemi.
ICD10: nenalezeno
Ontologie Definice: Uspoˇr´ad´an´ı koncept˚ u, pro kter´e lze vytvoˇrit racion´aln´ı argumenty. Zdroj: http://skmtglossary.org/search.aspx? term_id=1319&SearchExp=ontology
Synonyma: Gateway (pˇrejat´e slovo z angl.) Zdroj: http://skmtglossary.org/search.aspx? term_id=1006&SearchExp=gateway SNOMED CT: nenalezeno MeSH: nenalezeno
SNOMED CT: nenalezeno
ICD10: nenalezeno
MeSH: nenalezeno
Reference
ICD10: nenalezeno
Dom´ ena Definice: Vymezen´a oblast z´ ajmu nebo specializace. Zdroj: http://skmtglossary.org/search.aspx? term_id=1154&SearchExp=domain SNOMED CT: nenalezeno MeSH: nenalezeno ICD10: nenalezeno
[1] Kolektiv autor˚ u DASTA. V´ yvoj DASTA — Z´ akladn´ı informace — DASTA [Internet]. [citov´ an 5. duben 2014]. Dostupn´ e z: http://www.dastacr.cz/info-1.html [2] Blobel B. Application of the Component Paradigm for Analysis and Design of Advanced Health System Architectures. International Journal of Medical Informatics. 18. ˇ cervenec 2000;60:281–301. [3] Oemig F. PhD Thesis: Entwicklung einer ontologiebasierten Architektur zur Sicherung semantischer Interoperabilit¨ at zwischen Kommunikationsstandards im Gesundheitswesen [Internet]. [citov´ ano 26. ˇ cervenec 2012]. Dostupn´ e z: http://www.oemig.de/Frank/phd-thesis.htm [4] Oemig F, Blobel B. A Communication Standards Ontology Using Basic Formal Ontologies. Studies in Health Technology and Informatics. 2010. s. 105–13.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Schlenker A., Reimer M. – Velk´a data v nemocniˇcn´ıch informaˇcn´ıch syst´emech z pohledu bezpeˇcnosti
Velk´ a data v nemocniˇ cn´ıch informaˇ cn´ıch syst´ emech z pohledu bezpeˇ cnosti Anna Schlenker1,2 , Michal Reimer2 1 2
´ ˇ a republika Ustav hygieny a epidemiologie 1.LF a VFN, 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova v Praze, Cesk´
ˇ e vysok´e uˇcen´ı technick´e v Praze, Kladno, Katedra biomedic´ınsk´e informatiky, Fakulta biomedic´ınsk´eho inˇzen´yrstv´ı, Cesk´ ˇ a republika Cesk´
Kontakt: Anna Schlenker ´ ˇ a republika Ustav hygieny a epidemiologie 1.LF a VFN, Univerzita Karlova v Praze, Cesk´ Adresa: Studniˇ ckova 7, 128 00 Praha 2 E–mail:
[email protected]
C´ıle v´ yzkumu
Souˇ casn´ y stav pozn´ an´ı
V oblasti zdravotnictv´ı bylo v pr˚ ubˇehu let nahromadˇen´e velk´e mnoˇzstv´ı dat. V dneˇsn´ı dobˇe je poˇr´ad vˇetˇsina pacientsk´e dokumentace v pap´ırov´e podobˇe. Postupnˇe se vˇsak pˇrech´ az´ı k dokumentaci elektronick´e a doch´az´ı tak i ke konverzi pap´ırov´e formy do elektronick´e podoby. Zde se ovˇsem dost´ av´ ame ke zpracov´av´an´ı velk´eho objemu dat z elektronick´ ych zdravotn´ıch z´aznam˚ u, kter´ ych neust´ ale pˇrib´ yv´ a [1]. Velk´a data m˚ uˇzeme definovat pomoc´ı definice 4V“ [2]: ”
T´ema bezpeˇcnosti je v oblasti nemocniˇcn´ıch informaˇcn´ıch syst´em˚ u velmi aktu´aln´ı. Je dobˇre, ˇze jsou souˇcasn´e ordinace bez poˇc´ıtaˇce a bez informaˇcn´ıho syst´emu v nˇem jiˇz vz´acnost´ı. Na stranu druhou se vˇsak hodnˇe lid´ı zaˇc´ın´a zaj´ımat o to, kdo m´a pˇr´ıstup do tˇechto informaˇcn´ıch syst´em˚ u a k informac´ım v nich uloˇzen´ ym [4]. V souˇcasn´e dobˇe klademe velk´ y d˚ uraz na to, ˇze kaˇzd´ y uˇzivatel informaˇcn´ıho syst´emu mus´ı m´ıt sv´e vlastn´ı pˇrihlaˇsovac´ı u ´daje. Dalˇs´ı d˚ uleˇzitou vˇec´ı je nastaven´ı rozd´ıln´ ych pr´av v syst´emu pro r˚ uzn´e uˇzivatele (l´ekaˇr, zdravotn´ı sestra, laborant, radiolog, technick´ y pracovn´ık atd.). Pro uˇzivatele nemocniˇcn´ıch informaˇcn´ıch syst´em˚ u se poˇr´adaj´ı ˇskolen´ı, kde se zdravotniˇct´ı pracovn´ıci seznamuj´ı s potˇrebou multifaktorov´eho zabezpeˇcen´ı citliv´ ych pacientsk´ ych dat [4]. Hlavn´ım d˚ uvodem je to, aby si person´al v nemocnic´ıch uvˇedomil, ˇze zad´an´ı hesla do informaˇcn´ıho syst´emu nen´ı pouze nˇeco, co ho zdrˇzuje od pr´ace, ale tak´e nˇeco, co ho m˚ uˇze ochr´anit. D´ale pak nauˇcit uˇzivatele, ˇze nemocniˇcn´ı informaˇcn´ı syst´em nen´ı pouze obtˇeˇzuj´ıc´ı software, kter´ y jim neumoˇzn´ı pokraˇcovat v pr´aci bez vyplnˇen´ı nˇekter´ ych pol´ı, ale uk´azat jim, ˇze tyto poloˇzky m˚ uˇzou b´ yt d˚ uleˇzit´e, a proto bez jejich vyplnˇen´ı nelze z´aznam uloˇzit [4].
1. V = volume (objem) znamen´ a, ˇze se objem dat exponenci´alnˇe zvyˇsuje; 2. V = velocity (rychlost) znamen´ a, ˇze je na nˇekter´ ych pracoviˇst´ıch potˇreba rychle zpracov´avat velk´e mnoˇzstv´ı dat, kter´ a neust´ ale vznikaj´ı. 3. V = variety (promˇenlivost) znamen´ a, ˇze se kromˇe strukturovan´ ych dat zpracov´ av´ a tak´e nestrukturovan´ y text a nebo r˚ uzn´e typy multimedi´ aln´ıch dat.
4. V = veracity (pravdivost); znamen´ a, ˇze data nemaj´ı pˇresnˇe urˇcenou m´ıru pravdivosti kv˚ uli jecasn´ y stav bezpeˇ cnosti v nemocniˇ cn´ıch jich nekonzistenci, nekompletnosti a mnohokr´at Souˇ informaˇ cn´ıch syst´ emech i dvojv´ yznamovosti. N´asledn´a anal´ yza nasb´ıran´ ych velk´ ych dat m˚ uˇze tak´e zlepˇsit kvalitu zdravotn´ı p´eˇce. M˚ uˇze pomoct pˇri podpoˇre rozhodov´an´ı, pˇri anal´ yze rizik ˇci pˇri n´ akladov´e anal´ yze ve zdravotnictv´ı. Dalˇs´ı moˇznost´ı je pouˇzit´ı pˇri v´ yvoji l´ekaˇrsk´ ych doporuˇcen´ı [1]. Statistick´ a anal´ yza m˚ uˇze tak´e pomoct s lepˇs´ı interpretac´ı dat shrom´ aˇzdˇen´ ych v nemocniˇcn´ıch informaˇcn´ıch syst´emech [3]. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
V dneˇsn´ı dobˇe jiˇz vˇetˇsina nemocniˇcn´ıch informaˇcn´ıch syst´em˚ u pamatuje na bezpeˇcnost a nab´ız´ı moˇznost nastaven´ı r˚ uzn´ ych pr´av pro r˚ uzn´e uˇzivatele. Bohuˇzel vˇsak vˇetˇsina informaˇcn´ıch syst´em˚ u ovˇeˇruje uˇzivatele pouze jednou autentizaˇcn´ı metodou, typicky heslem. Toto heslo nav´ıc nemus´ı splˇ novat poˇzadavky potˇrebn´e napˇr´ıklad pro odolnost proti slovn´ıkov´ ym
39
40
Schlenker A., Reimer M. – Velk´a data v nemocniˇcn´ıch informaˇcn´ıch syst´emech z pohledu bezpeˇcnosti
u ´tok˚ um. Sami uˇzivatel´e by proto mˇeli pamatovat na to, ˇze by jako heslo nemˇeli pouˇz´ıvat jm´ena sv´ ych partner˚ u, dˇet´ı, dom´ac´ıch mazl´ıˇck˚ u ˇci jin´ a plnov´ yznamov´ a slova. Obecnˇe by heslo mˇelo b´ yt dostateˇcnˇe dlouh´e (nejm´enˇe 8 znak˚ u), mˇelo by obsahovat velk´ a i mal´ a p´ısmena, ˇc´ıslice a speci´aln´ı znaky. Za bezpeˇcn´e heslo se povaˇzuje heslo, kter´e uˇzivatel nikomu nesdˇel´ı (ani manˇzelovi ˇci manˇzelce) a nikam si nezap´ıˇse. Pro u ´toˇcn´ıky nen´ı nic jednoduˇsˇs´ıho neˇz opsat heslo, kter´e m´a uˇzivatel zapsan´e v di´ aˇri ˇci na pap´ırku nalepen´em Obr´ azek 1: Doba trv´ an´ı stisku kl´ avesy a doba mezi jednotna monitoru [4]. liv´ ymi stisky.
Dalˇs´ı chybou vˇetˇsiny nemocniˇcn´ıch informaˇcn´ıch syst´em˚ u je, ˇze nepouˇz´ıvaj´ı automatick´e odhlaˇsov´an´ı uˇzivatel˚ u. D˚ uvodem je pravdˇepodobnˇe ztr´ ata ˇcasu pro zdravotnick´ y person´al, kter´ y se mus´ı neust´ ale do syst´emu pˇrihlaˇsovat. Na druhou stranu by si ale uˇzivatel´e nemocniˇcn´ıch informaˇcn´ıch syst´em˚ u mˇeli uvˇedomit, ˇze opuˇstˇen´ı poˇc´ıtaˇce s pˇrihl´ aˇsen´ ym informaˇcn´ım syst´emem m˚ uˇze zp˚ usobit ztr´atu ˇci zmˇenu citliv´ ych pacientsk´ ych dat uloˇzen´ ych v dan´em syst´emu [4].
Pro lepˇs´ı a pˇrehlednˇejˇs´ı zpracov´an´ı dat jsme aplikaci rozˇs´ıˇrili o druh´ y listview (viz Obr´azek 2), kde m´ame moˇznost zobrazen´ı druh´eho z´aznamu. Aplikace disponuje tak´e postrann´ım panelem (viz Obr´azek 2), kde se po oznaˇcen´ı konkr´etn´ıch ˇr´adk˚ u v obou z´aznamech automaticky vypoˇc´ıtaj´ı parametry dynamiky stisku poˇc´ıtaˇcov´ ych kl´aves pro jednotliv´e uˇzivatele. Aplikace umoˇzn ˇuje takt´eˇz naˇcten´ı dˇr´ıve nasn´ıman´ ych a uloˇzen´ ych z´aznam˚ u a d´av´a n´am t´ım moˇznost i zpˇetn´e anal´ yzy [5].
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı Pro zlepˇsen´ı zabezpeˇcen´ı v oblasti nemocniˇcn´ıch informaˇcn´ıch syst´em˚ u jsme navrhli aplikaci, kter´ a pouˇz´ıv´ a multifaktorovou autentizaci, kter´ a kombinuje znalostn´ı faktor a biometrick´ y faktor. Jako biometrick´ y faktor jsme pouˇzili behavior´ aln´ı biometrickou charakteristiku nazvanou dynamika stisku poˇc´ıtaˇcov´ ych kl´ aves. Dynamika stisku poˇc´ıtaˇcov´ ych kl´ aves pˇresnˇe popisuje styl psan´ı dan´eho uˇzivatele na kl´avesnici, a to d´ıky sn´ım´ an´ı ˇcas˚ u, kdy byla konkr´etn´ı kl´avesa stlaˇcena a kdy uvolnˇena. Aplikaci lze pouˇz´ıt n´ asleduj´ıc´ımi dvˇema zp˚ usoby: Obr´ azek 2: Koneˇcn´ a verze aplikace pro sn´ım´ an´ı dynamiky stisku poˇc´ıtaˇcov´ ych kl´ aves.
1. Jako aplikaci pro statick´e ovˇeˇrov´ an´ı uˇzivatel˚ u (zad´an´ı pˇrihlaˇsovac´ıho jm´ena + hesla + sn´ım´an´ı Aplikace byla implementov´ana v jazyce C] za pomoc´ı dynamiky stisku poˇc´ıtaˇcov´ ych kl´ aves pˇri psan´ı Microsoft Visual Studio Express 2012. Nasn´ıman´a data se uˇzivatelsk´eho jm´ena a hesla). ukl´adaj´ı do souboru CSV.
Podˇ ekov´ an´ı
2. Jako aplikaci pro kontinu´ aln´ı ovˇeˇrov´ an´ı identity uˇzivatel˚ u (ovˇeˇrov´ an´ı, zda uˇzivatel, kter´ y pracuje Tato pr´ace byla podpoˇrena projektem SVV-2015s informaˇcn´ım syst´emem, je uˇzivatel, kter´ y je do 260158 Univerzity Karlovy v Praze. syst´emu pˇrihl´aˇsen´ y).
Kl´ıˇ cov´ a slova
Nejvˇetˇs´ı v´ yhodou t´eto aplikace je sbˇer dat pˇr´ımo z operaˇcn´ıho syst´emu, tj. bez zpoˇzdˇen´ı. Aplikace zaznaVelk´ a data men´av´a k´od dan´e kl´avesy, n´ azev t´eto kl´ avesy a ˇcasy, kdy byla kl´avesa stlaˇcena a kdy uvolnˇena. Definice: Velk´a data lze definovat pomoc´ı tzv. 4V“ de” Naˇse aplikace umoˇzn ˇuje tak´e automatickou anal´ yzu finice, kde V znamen´a objem (volume), rychlost (vesn´ıman´ ych dat. Pro u ´ˇcely t´eto anal´ yzy aplikace sama locity), rozmanitost (variabilitu) a pravdivost (verapoˇc´ıt´a ˇcasov´ y vektor, kter´ y se skl´ ad´ a z dob trv´ an´ı stisku city). jednotliv´ ych kl´aves a z dob mezi jednotliv´ ymi stisky (viz Zdroj: [1] Obr´azek 1). S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Schlenker A., Reimer M. – Velk´a data v nemocniˇcn´ıch informaˇcn´ıch syst´emech z pohledu bezpeˇcnosti
SNOMED CT: nenalezeno
Zdroj: [6]
MeSH: nenalezeno
SNOMED CT: nenalezeno
ICD10: nenalezeno
MeSH: nenalezeno
Informaˇ cn´ı syst´ em
ICD10: nenalezeno
Definice: Integrovan´a sada soubor˚ u, postup˚ u, a zaˇr´ızen´ı Dynamika stisku poˇ c´ıtaˇ cov´ ych kl´ aves pro skladov´ an´ı, manipulaci a vyhled´ av´ an´ı informac´ı. Definice: Detailn´ı informace o ˇcasech, kter´e v pr˚ ubˇehu Zdroj: http://www.nlm.nih.gov/cgi/mesh/2015/MB_ psan´ı uˇzivatele na kl´avesnici pˇresnˇe popisuj´ı, kdy cgi?mode=&term=Information+Systems &field= byla kter´a kl´avesa stlaˇcena a kdy uvolnˇena. entry#TreeL01.700.508.300 Synonyma: Dynamika psan´ı na kl´avesnici SNOMED CT: 706593004 MeSH: D007256
Zdroj: [7]
ICD10: nenalezeno
SNOMED CT: nenalezeno
Nemocniˇ cn´ı informaˇ cn´ı syst´ em
MeSH: nenalezeno ICD10: nenalezeno
Definice: Integrovan´ y, poˇc´ıtaˇcem podporovan´ y syst´em urˇcen´ y k ukl´ ad´ an´ı, manipulaci a z´ısk´ av´ an´ı informac´ı t´ ykaj´ıc´ıch se administrativn´ıch a klinick´ ych hledisek poskytovan´ ych zdravotnick´ ych sluˇzeb v r´amci nemocnice. Synonyma: NIS Zdroj: http://www.nlm.nih.gov/cgi/mesh/2015/ MB_cgi?mode=&term=Hospital+Information +Systems&field=entry#TreeL01.700.508.300. 408 SNOMED CT: 462944003 MeSH: D006751 ICD10: nenalezeno
Multifaktorov´ a autentizace Definice: Bezpeˇcnostn´ı syst´em, v nˇemˇz se pouˇz´ıv´a v´ıce neˇz jedna forma autentizace k ovˇeˇren´ı opr´avnˇenosti operace.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Reference [1] Raghupathi W., Raghupathi V.: Big data analytics in healthcare: promise and potential. Health Information Science and Systems 2014, 2(3). doi:10.1186/2047-2501-2-3 [2] Schlenker A., Bohunˇ c´ ak A.: Keystroke Dynamics for Security Enhancement in Hospital Information Systems. International Journal on Biomedicine and Healthcare 2015; 3(1):41–44 [3] Kalina J.: Statistical Challenges of Big Data Analysis in Medicine. International Journal on Biomedicine and Healthcare 2015; 3(1):24–27 [4] Schlenker A.: Multifactor Data Security in Information Systems in Health Care. International Journal on Biomedicine and Healthcare 2014; 2(1):25–27 [5] Reimer M., Schlenker A. Ovˇ eˇrov´ an´ı identity na z´ akladˇ e kontinu´ aln´ıho sn´ım´ an´ı dynamiky stisku poˇ c´ıtaˇ cov´ ych kl´ aves. ˇ Kladno, 2015. Bachelor thesis. Cesk´ e vysok´ e uˇ cen´ı technick´ e. [6] Badr Y., Chbeir R., Abraham A., Hassanien A.-E. (Eds.): Emergent Web Intelligence: Advanced Semantic Technologies, 1st Edition., 2010, XVI, 544, p.345 [7] Bergadano F., Gunetti D., Picardi C.: User authentication through Keystroke Dynamics. ACM Transactions on Information and System Security (TISSEC), 2002;5(4): 367–397
41
42
Siˇcov´a K. a kol. – Sekund´arn´ı katarakta u pacient˚ u po implantaci multifok´aln´ıch nitrooˇcn´ıch ˇcoˇcek
Sekund´ arn´ı katarakta u pacient˚ u po implantaci multifok´ aln´ıch nitrooˇ cn´ıch ˇ coˇ cek Krist´ına Siˇ cov´ a1 , Petr V´ yborn´ y1 , Jiˇr´ı Paˇsta1 1
´ redn´ı Vojensk´ ˇ a republika Oˇcn´ı klinika, Ustˇ a nemocnice – Vojensk´ a fakultn´ı nemocnice, Praha, Cesk´
Kontakt: Krist´ına Siˇ cov´ a ´ Oˇ cn´ı klinika UVN Praha Adresa: U Vojensk´ e nemocnice 1200, 16902 Praha 6 E–mail:
[email protected]
C´ıle v´ yzkumu
retrospektivnˇe vyhodnotit zaznamenan´a data, tak i prospektivnˇe, s vyuˇzit´ım modern´ıch objektivn´ıch zobrazoV posledn´ıch letech roste ve vyspˇel´ ych zem´ıch poˇcet vac´ıch metod, sledovat zkoumanou problematiku. prov´adˇen´ ych operac´ı zamˇeˇren´ ych na pacienty ve stˇredn´ım a vyˇsˇs´ım vˇeku, kteˇr´ı pˇrich´ azej´ı do ordinace oˇcn´ıho l´ekaˇre casn´ y stav pozn´ an´ı s poˇzadavkem zbavit se z´ avislosti na br´ ylov´e korekci anebo Souˇ ji alespoˇ n v´ yraznˇe omezit. Tuto moˇznost nab´ız´ı refrakˇcn´ı chirurgie d´ıky ˇsirok´emu V souˇcasn´e dobˇe je zn´amo, ˇze za rozvoj opacifikace spektru nitrooˇcn´ıch ˇcoˇckov´ ych implant´ at˚ u. zadn´ıho pouzdra ˇcoˇcky je zodpovˇedn´a proliferace a miˇ e republice jsou v pr˚ V Cesk´ ubˇehu posledn´ıch let grace bunˇek ˇcoˇckov´eho epitelu (LEC- lens epithelial cells) nejˇcastˇeji implantovan´e multifok´ aln´ı difrakˇcn´ı nitrooˇcn´ı zejm´ena k v oblasti ekv´atoru ˇcoˇcky. ˇcoˇcky. N´asledn´a katarakta m´a dvˇe patogenetick´e formy: Jedn´a se o z´akrok n´ aroˇcn´ y nejenom na kvalitu prove- fibr´ozn´ı zmˇeny pˇredn´ıho a zadn´ıho listu pouzdra ˇcoˇcky den´ı, erudici operat´era, spolupr´ aci pacienta, ale z´ aroveˇ n a regeneraˇcn´ı ve smyslu novotvoˇren´ ych ˇcoˇckov´ ych hmot. i z´akrok zat´ıˇzen´ y finanˇcn´ım doplatkem ze strany pacienta. Na zrakovou ostrost maj´ı vliv pouze opacity nach´azej´ıc´ı Proto lze pˇredpokl´ adat, ˇze pacient od operace se v optick´e ose vidˇen´ı. vyˇzaduje co nejlepˇs´ı v´ ysledky a jak´ akoliv pooperaˇcn´ı komKontakt optick´e ˇc´asti IOL (intraocular lens) s buˇ nkami plikace ˇci zhorˇsen´a zrakov´ a ostrost se st´ av´ a v´ yznamn´ ym ˇcoˇckov´eho epitelu pˇredn´ıho pouzdra vede k jejich diferenprobl´emem. ciaci na myofibroblasty, n´aslednˇe k fibrotizaci a rozvoji Nejˇcastˇejˇs´ı pozdn´ı komplikac´ı po jinak nekomplikoACO (anterior capsule cataract), jejich migrac´ı na zadn´ı van´e operaci je rozvoj opacifikace zadn´ıho pouzdra (PCO list doch´az´ı ke vzniku PCO. Na vzniku regeneraˇcn´ıho – posterior capsule opacification) implantovan´e nitrooˇcn´ı typu sekund´arn´ı katarakty se pod´ıl´ı proliferace a migrace ˇcoˇcky, sniˇzuj´ıc´ı zrakovou ostrost pacienta, kter´ a m˚ uˇze v´est LEC, kter´e z˚ ustaly pˇri operaci v oblasti ekv´atoru (tzv. aˇz k rozvoji tupozrakosti, ˇcili amblyopie. E-buˇ nky). Jednak se vytv´aˇr´ı nov´e ˇcoˇckov´e hmoty pod´el Tato problematika nab´ yv´ a v´ yznamu u pacient˚ u po ekv´atoru (Sommering˚ uv prstenec), jednak migruj´ı mezi operaci s implantac´ı pr´emiov´e nitrooˇcn´ı ˇcoˇcky se sloˇzitou zadn´ı list pouzdra a IOL, kde vytv´aˇr´ı tzv. Elschnigovy optikou, kde kaˇzd´a eventu´ aln´ı komplikace v´ yraznˇe sniˇzuje perly [1]. ˇci znehodnocuje moˇznost vyuˇzit´ı multifok´ aln´ı optiky. V prevenci vzniku sekund´arn´ı katarakty byl prok´az´an C´ılem naˇseho v´ yzkumu je vyhodnotit, jak´e faktory vliv tvaru a materi´alu IOL, peroperaˇcn´ı i poovedou k rozvoji sekund´ arn´ı katarakty, zda jej´ı rozperaˇ c n´ ı l´eˇcby, vliv samotn´e chirurgick´e techniky operace. voj ovlivˇ nuje pouze zp˚ usob proveden´ı operace, pouˇzit´a Souˇ c asn´ ym uzn´avan´ ym preventivn´ım postupem je imoperaˇcn´ı technika a instrument´ arium anebo pˇrisp´ıvaj´ı plantace IOL do pouzdra p˚ uvodn´ı ˇcoˇcky, pouˇzit´ı IOL i faktory ze strany pacienta, a to vˇek, pohlav´ı, dalˇs´ı celkov´e s ostr´ y mi zadn´ ımi hranami, extenzivn´ı ˇciˇstˇen´ı kapsuly. onemocnˇen´ı pacienta, pˇr´ıtomnost ˇci nepˇr´ıtomnost ˇsed´eho Pˇ r esto ani pot´ e nen´ ı vznik komplikac´ ı spojen´ ych s rozz´akalu pˇredoperaˇcnˇe, jeho typ, ˇci v´ yˇse pˇredoperaˇcn´ı divojem sekund´ a rn´ ı katarakty v´ y jimeˇ c n´ y [2]. optrick´e refrakˇcn´ı vady. Po prvn´ım roce od operace je sekund´arn´ı kataraktou Na naˇsem pracoviˇsti prov´ ad´ıme refrakˇcn´ı z´ akroky s implantac´ı pr´emiov´ ych nitrooˇcn´ıch ˇcoˇcek jiˇz od roku 2005, postiˇzeno kolem 11% pacient˚ u (v naˇsem souboru 1,56%), ykonu aˇz 28% pacient˚ u [3], (v naˇsem a proto disponujeme pomˇernˇe velk´ ym souborem pravi- po pˇeti letech po v´ delnˇe sledovan´ ych pacient˚ u. Tento fakt n´ am umoˇzn ˇuje jak souboru 2,34%) (Obr´azek 1). S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Siˇcov´a K. a kol. – Sekund´arn´ı katarakta u pacient˚ u po implantaci multifok´aln´ıch nitrooˇcn´ıch ˇcoˇcek
reoretin´aln´ı chirurgii, ale i ve zdravotnictv´ı jako takov´em, pˇri posouzen´ı ekonomick´e efektivity jednotliv´ ych postup˚ u. Statistick´e vyhodnocen´ı v´ ysledk˚ u bude urˇcitˇe zaj´ımav´ ym pˇr´ınosem nejenom pro oˇcn´ı l´ekaˇre, ale d´ıky pohledu na pacienta jako takov´eho, vˇcetnˇe jeho intern´ıch onemocnˇen´ı, vˇeku, pohlav´ı, najdou v´ ysledky uplatnˇen´ı i v ostatn´ıch odvˇetv´ıch medic´ıny, biomedic´ınsk´e statistiky a moˇzn´a pomohou i pacient˚ um zvaˇzuj´ıc´ım podstoupen´ı operace.
Podˇ ekov´ an´ı Obr´ azek 1: Procentu´ aln´ı vyj´ adˇren´ı incidence sekund´ arn´ı katarakty po implantaci MIOL v pr˚ ubˇehu let.
Pr˚ umˇern´a variace je od 10% do 40% tˇri aˇz pˇet let od operace [4]. U nˇekter´ ych typ˚ u IOL je to pouze 5% [5].
Tato pr´ace byla podpoˇrena projektem SVV-2015260158 Univerzity Karlovy v Praze.
Kl´ıˇ cov´ a slova
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı
Opacifikace zadn´ıho pouzdra ˇ coˇ cky
Sekund´arn´ı katarakta, i mal´eho rozsahu, m˚ uˇze v´est k v´ yrazn´emu sn´ıˇzen´ı zrakov´e ostrosti u pacient˚ u s multifok´aln´ı nitrooˇcn´ı ˇcoˇckou. Horˇsen´ı vidˇen´ı se projev´ı zejm´ena na bl´ızkou vzd´ alenost. Tato mal´ a, u monofok´aln´ıch IOL nesignifikantn´ı PCO se m˚ uˇze rozvinout jiˇz p˚ ul roku od operace [6]. Nejˇcastˇejˇs´ı terapi´ı z˚ ust´ av´ a NdYAG kapsulotomie. Tato nese sebou, aˇc mal´e, poˇr´ ad riziko rozvoje trhlin v s´ıtnici, odchl´ıpen´ı s´ıtnice, vzniku sklivcov´ ych z´ akalk˚ u, posunu ˇci decentrace nitrooˇcn´ı ˇcoˇcky a t´ım zhorˇsen´ı jej´ı funkce. T´ım se i pˇres preciznˇe proveden´e mˇeˇren´ı s´ıly nitrooˇcn´ı ˇcoˇcky, nekomplikovan´e proveden´ı operace a ze zaˇc´atku klidn´ y pooperaˇcn´ı pr˚ ubˇeh mohou rozvinout dalekos´ahl´e komplikace a vyˇz´ adat si dalˇs´ı chirurgick´ y z´ asah, vˇcetnˇe vitreoretin´aln´ı (s´ıtnicov´e) chirurgie, hospitalizaci pacienta. N´asledkem m˚ uˇze b´ yt trval´e sn´ıˇzen´ı zrakov´e ostrosti pacienta, jeho invalidizace, omezen´ı v pracovn´ım ˇci soukrom´em ˇzivotˇe a t´ım i z toho plynouc´ı finanˇcn´ı n´asledky jak ze strany pacienta, tak na stranˇe zdravotnictv´ı. D´ıky jiˇz zm´ınˇen´emu rostouc´ımu z´ ajm˚ u jak fakochirurg˚ u, tak pacient˚ u o multifok´ aln´ı implant´aty se aktu´alnost t´eto problematiky zvyˇsuje. C´ılem oˇcn´ıch chirurg˚ u je sn´ıˇzit riziko vzniku sekund´arn´ı katarakty na minimum a zachovat po co nejdelˇs´ı dobu funkci tˇechto implant´ at˚ u. V prozat´ımn´ıch v´ ysledc´ıch naˇseho v´ yzkumu vid´ıme niˇzˇs´ı incidenci sekund´ arn´ı katarakty, neˇz je publikov´ano ve svˇetov´e literatuˇre. Tento v´ ysledek dle naˇseho n´ azoru ovlivˇ nuj´ı faktory, nez´avisl´e na materi´ alu a typu nitrooˇcn´ıho implant´atu, a to n´ami vypracovan´ y syst´em peˇcliv´eho v´ ybˇeru vhodn´ ych kandid´at˚ u k operaci, protokol proveden´ı operace a pooperaˇcn´ı terapie, jakoˇz i syst´em pooperaˇcn´ıch kontrol, vˇcetnˇe zaveden´ ych pˇr´ısn´ ych reˇzimov´ ych opatˇren´ı. Dovol´ıme si tvrdit, ˇze v´ ysledky naˇseho v´ yzkumu mohou naj´ıt svoje uplatnˇen´ı jak v kataraktov´e chirurgii, vit-
Synonyma: N´asledn´a katarakta, sekund´arn´ı katarakta, sekund´arn´ı ˇsed´ y z´akal
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Definice: Zakalen´ı ˇcoˇckov´eho pouzdra.
ˇ Vlk F., Lexikon oˇcn´ıho Zdroj: Vlkov´a E., Pitrov´a S., l´ekaˇrstv´ı, 2008; 355 SNOMED CT: 410567004 MeSH: D058442 ICD10: H264
Katarakta Definice: Z´akal ˇcoˇcky v oku, kter´ y vede k rozptylu svˇetla vstupuj´ıc´ıho do oka a s postupuj´ıc´ım onemocnˇen´ım k postupn´emu zhorˇsov´an´ı vidˇen´ı. ˇ y z´akal Synonyma: Sed´ ˇ Vlk F., Lexikon oˇcn´ıho Zdroj: Vlkov´a E., Pitrov´a S., l´ekaˇrstv´ı, 2008; 212-223 SNOMED CT: 193570009 MeSH: D002386 ICD10: H25
Nitrooˇ cn´ı ˇ coˇ cka Definice: Umˇel´a ˇcoˇcka urˇcena k trval´emu uloˇzen´ı do oka. ˇ Vlk F., Lexikon oˇcn´ıho Zdroj: Vlkov´a E., Pitrov´a S., l´ekaˇrstv´ı, 2008; 82-85 SNOMED CT: 385468004 MeSH: D007910 ICD10: Z961
43
44
Siˇcov´a K. a kol. – Sekund´arn´ı katarakta u pacient˚ u po implantaci multifok´aln´ıch nitrooˇcn´ıch ˇcoˇcek
Multifok´ aln´ı nitrooˇ cn´ı ˇ coˇ cka
SNOMED CT: 85785001
Definice: Nitrooˇcn´ı ˇcoˇcka, kter´ a dok´ aˇze zaostˇrit na MeSH: D064727 s´ıtnici oka svˇeteln´e paprsky pˇrich´ azej´ıc´ı z bl´ızka ICD10: nenalezeno i z d´alky. ˇ Vlk F., Lexikon oˇcn´ıho Zdroj: Vlkov´a E., Pitrov´ a S., l´ekaˇrstv´ı, 2008; 82-85 SNOMED CT: 313236002 MeSH: nenalezeno ICD10: Z961
NdYAG kapsulotomie Definice: Chirurgick´e protˇet´ı zadn´ıho pouzdra ˇcoˇcky vedouc´ı k odstranˇen´ı druhotn´e katarakty neod´ ymov´ ym laserem. ˇ Vlk F., Lexikon oˇcn´ıho Zdroj: Vlkov´a E., Pitrov´ a S., l´ekaˇrstv´ı, 2008; 82-85
Reference [1] Sacu S., Menapace R., Findl O. et al. Influence of optic Eege design and anterior capsule polishing on posterior capsule fibrosis. Journal of cataract and refractive surgery, 2004; 30: 658-662 [2] Krajˇ cov´ a P., Chynoransk´ y M., Strmeˇ n P. Opacifik´ acia zadn´ eho p´ uzdra ˇsoˇsovky po implant´ acii rˆ oznych typov umel´ ych vn´ utrooˇ cn´ ych ˇsoˇsoviek - II. ˇ cast’ : rˆ ozne peroperaˇ cn´ e n´ alezy. ˇ Cesk´ a a slovensk´ a oftalmologie, 2008; 64: 13-15 [3] Bertelmann E., Kojetinsky C. Posterior capsule opacification. Current opinion in Ophthalmology, 2001; 12: 35-40 [4] Pandey S.K., Apple D.J., Wener L. et al. Posterior capsule opacification: A review of ethiopathogenesis. Experimental and clinical studies and factors for prevention. Ophthalmology, 2004;52 : 99-112 [5] Hayashi K., Hayashi H., Posterior capsule opacification in the presence of an intraocular lens with a sharp versus rounded optic edge, Ophthalmology, 2005; 112 : 1550-1556 [6] Larkin H., PCO and premium lens, Eurotimes, 2012; 17(5): 4-6
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
45
Slov´ak D., Zv´arov´a J. – DNA v biomedic´ınsk´ych aplikac´ıch
DNA v biomedic´ınsk´ ych aplikac´ıch Dalibor Slov´ ak1,2 , Jana Zv´ arov´ a1 1
´ ˇ a republika Ustav hygieny a epidemiologie, 1. LF UK, Praha, Cesk´ 2
´ ˇ Ustav zdravotnick´ych informac´ı a statistiky CR
Kontakt: Dalibor Slov´ ak ´ ˇ Ustav zdravotnick´ ych informac´ı a statistiky CR Adresa: Palack´ eho n´ am. 4, 128 01 Praha 2 E–mail:
[email protected]
C´ıle v´ yzkumu Vyuˇzit´ı DNA je dnes v odborn´e i laick´e veˇrejnosti bˇeˇznˇe pˇrij´ımanou prax´ı. Vyuˇz´ıv´ a se v mnoha oblastech kriminalistiky: k identifikaci pachatel˚ u zloˇcinu, pˇri urˇcov´an´ı otcovstv´ı, k identifikaci obˇet´ı hromadn´ ych neˇstˇest´ı apod. S klesaj´ıc´ı cenou anal´ yz se v posledn´ı dobˇe rozm´ah´a pouˇzit´ı DNA anal´ yzy tak´e v mnoha dalˇs´ıch smˇerech: zkoum´an´ı historick´ ych ostatk˚ u, sestavov´ an´ı rodokmen˚ u a hled´an´ı pˇr´ıbuzensk´ ych vazeb a v neposledn´ı ˇradˇe tak´e v biomedic´ınsk´ ych aplikac´ıch.
Souˇ casn´ y stav pozn´ an´ı Prvn´ım pˇr´ıkladem vyuˇzit´ı v medic´ınsk´ ych vˇed´ach je Huntingtonova chorea. Expanze v poˇctu trinukleotidov´ ych repetitivn´ıch sekvenc´ı u Huntingtonovy chorey zvyˇsuje pravdˇepodobnost v´ yskytu nevyv´ aˇzen´eho psychick´eho stavu. Pˇri z´ avaˇzn´e trestn´e ˇcinnosti je moˇzn´e posuzovat pˇr´ıˇcetnost jedince a schopnost jeho sebeovl´ad´an´ı. Jako stav mimo normu se ud´ av´ a poˇcet nad 35 repetitivn´ıch opakov´an´ı. V ˇcesk´e populaci je vrozen´ y sklon ke zv´ yˇsen´emu sr´aˇzen´ı krve (trombofilie) pomˇernˇe vysoce rozˇs´ıˇren – postihuje pˇribliˇznˇe 8 % obyvatel. Nejv´ıce rizikovou skupinou jsou ˇzeny uˇz´ıvaj´ıc´ı bˇeˇznou hormon´ aln´ı antikoncepci, u nichˇz je geneticky d´ an pˇredpoklad ke zv´ yˇsen´emu sr´aˇzen´ı krve. Pro prevenci a minimalizaci probl´em˚ u spojen´ ych se vznikem tromb´ ozy a tromboembolick´e nemoci je d˚ uleˇzit´a znalost genetick´ ych dispozic, nebot’ ohroˇzen´a osoba m˚ uˇze na z´ akladˇe znalosti sv´ ych dˇediˇcn´ ych vloh vhodnˇe pˇrizp˚ usobit svou ˇzivotospr´ avu a ˇzivotn´ı styl. Proto doch´az´ı k anal´ yze trombofiln´ıch mutac´ı i u d´arkyˇ n z´arodeˇcn´ ych bunˇek. Nejbˇeˇznˇejˇs´ı je faktor V Leiden (G1691A), zp˚ usobuj´ıc´ı rezistenci faktoru V k antikoagulaˇcn´ı aktivitˇe APC proteinu. Jednou z motivac´ı pro rozvoj personalizovan´e medic´ıny je to, ˇze u ´ˇcinek nˇekter´ ych l´ek˚ u m˚ uˇze b´ yt ovlivnˇen genetickou informac´ı pacienta, tedy ˇze odpovˇed’ organismu na l´eky m˚ uˇze b´ yt geneticky determinovan´ a. Pro l´ekaˇre S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
je d˚ uleˇzit´e prov´est test jeˇstˇe pˇred zah´ajen´ım l´eˇcby nebo pˇri u ´vaze o jej´ım rozˇs´ıˇren´ı ˇci zmˇenˇe. Jedn´ım z pˇr´ıklad˚ u je gen VKORC1, jenˇz m´a 24 zn´am´ ych alel. Dvˇe z nich (CYP2C9-2, CYP2C9-3) zvyˇsuj´ı antikoagulaˇcn´ı efekt warfarinu a sniˇzuj´ı denn´ı d´avky potˇrebn´e k udrˇzen´ı INR (International Normalized Ratio) v terapeutick´em rozsahu. V genu UGT1A1 je u nositel˚ u alely 28 pozorov´ana tˇeˇzk´a toxicita Irinotecanu, l´eku pouˇz´ıvan´eho zejm´ena pˇri l´eˇcbˇe metast´azuj´ıc´ıch n´ador˚ u tlust´eho stˇreva a rekta.
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı DNA m´a z hlediska testov´an´ı dvˇe velk´e v´ yhody. Lidsk´a DNA se skl´ad´a z pˇribliˇznˇe 23 000 gen˚ u, nˇekter´e z nich maj´ı ale v´ıce transkript˚ u, takˇze znak˚ u vyuˇziteln´ ych k odliˇsen´ı ˇ adn´ı dva lid´e na svˇetˇe nemaj´ı shodn´ je jeˇstˇe v´ıce. Z´ y cel´ y genom, dokonce ani jednovajeˇcn´a dvojˇcata. Druhou v´ yhodou DNA je jej´ı vˇsudypˇr´ıtomnost. DNA se vyskytuje v kaˇzd´e buˇ nce a z˚ ust´av´a na vˇsech m´ıstech, kde se pohybujeme. Tato v´ yhoda snadn´e dostupnosti je ovˇsem z´aroveˇ n tak´e nev´ yhodou. D´ıky citlivosti dneˇsn´ıch pˇr´ıstroj˚ u je moˇzno analyzovat i nˇekolik m´alo bunˇek. Pˇri anal´ yze takov´ ych vzork˚ u se vˇsak ˇcasto vyskytuj´ı smˇesn´e DNA vzorky nebo kontami´ ernˇe k tomu se zvyˇsuje d˚ nace neˇcistotami. Umˇ uleˇzitost zahrnut´ı vˇsech moˇzn´ ych stochastick´ ych jev˚ u, spojen´ ych s laboratorn´ım zpracov´an´ım, do v´ ypoˇctu. Dalˇs´ım probl´emem DNA anal´ yzy je jej´ı restrikce na vybranou ˇc´ast genomu. Ve forenzn´ı praxi nen´ı vyhodnocov´ana cel´a DNA, ale je sestavov´an tzv. DNA profil. Jedn´ım d˚ uvodem je cena: sekvenace genomu jednoho ˇclovˇeka jiˇz stoj´ı m´enˇe neˇz 1 milion USD a d´ale kles´a, nicm´enˇe pro rutinn´ı anal´ yzy se jedn´a st´ale o velmi vysokou cenu. Druh´ ym d˚ uvodem je citlivost u ´daj˚ u, kter´e lze z DNA z´ıskat. Ve sloˇzitˇejˇs´ıch pˇr´ıpadech proto nen´ı jednoznaˇcn´a identifikace moˇzn´a a ke slovu se dost´avaj´ı pravdˇepodobnostn´ı modely, zahrnuj´ıc´ı r˚ uzn´e mnoˇzstv´ı parametr˚ u. Tyto modely obvykle nˇejak´ ym zp˚ usobem
46
Slov´ak D., Zv´arov´a J. – DNA v biomedic´ınsk´ych aplikac´ıch
poˇc´ıtaj´ı pravdˇepodobnost pozorov´ an´ı dan´eho DNA profilu v populaci. Nˇekter´e pouˇz´ıvan´e modely mohou kv˚ uli jednoduchosti opom´ıjet parametry, kter´e mohou b´ yt pro v´ ysledek podstatn´e, nebo mohou tyto parametry zahrnovat do v´ ypoˇctu ˇspatn´ ym zp˚ usobem. Z toho vznikaj´ı chyby, kter´e mohou z´asadn´ım zp˚ usobem zmˇenit v´ ysledek cel´e DNA anal´ yzy. Jako memento je moˇzn´e uv´est pˇr´ıpad popsan´ y v ˇcl´anku [1]: Autoˇri pouˇzili smˇesn´ y vzorek DNA ze skuteˇcn´eho pˇr´ıpadu a pˇredloˇzili jej 17 zkuˇsen´ ym analytik˚ um pracuj´ıc´ım v akreditovan´e vl´ adn´ı laboratoˇri v USA, aniˇz by jim poskytli nˇejakou doplˇ nkovou informaci, kter´a by mohla vych´ ylit jejich rozhodnut´ı. Bylo dosaˇzeno zcela protich˚ udn´ ych z´avˇer˚ u o tom, zda DNA podezˇrel´e osoby koreluje se smˇesn´ ym vzorkem z m´ısta ˇcinu. Pouze jeden analytik souhlasil s p˚ uvodn´ım rozhodnut´ım v soudn´ım procesu, ˇze podezˇrel´ y nem˚ uˇze b´ yt vylouˇcen jako potenci´ aln´ı paˇ ri analytici prohl´ chatel. Ctyˇ asili, ˇze d˚ ukazn´ı evidence je nepr˚ ukazn´a, a zbyl´ ych 12, ˇze podezˇrel´eho lze jako pachatele vylouˇcit. N´asledn´ y pr˚ uzkum laboratoˇr´ı po cel´em svˇetˇe uk´azal, ˇze existuj´ı v´ yrazn´e nesrovnalosti v postupech doporuˇcen´ ych pro interpretaci vzork˚ u. Jak m˚ uˇzeme vidˇet, anal´ yza DNA m´ a tak´e sv´a rizika. Je tˇreba sjednotit terminologii, zlepˇsit a standardizovat pouˇz´ıvan´e pravdˇepodobnostn´ı modely a drˇzet se doporuˇcen´ ych postup˚ u, nicm´enˇe ani pˇri jejich dodrˇzen´ı nebude moˇzn´e zcela odstranit stochastickou povahu nˇekter´ ych jev˚ u. Proto je nutn´e anal´ yzu DNA vn´ımat jako d˚ uleˇzit´ y, ale nikoli neomyln´ y n´ astroj modern´ı vˇedy.
Anal´ yza DNA
Podˇ ekov´ an´ı
SNOMED CT: 234467004
Definice: Metoda forenzn´ıch vˇed vyuˇz´ıvaj´ıc´ı biologick´ y materi´al obsahuj´ıc´ı DNA. Zdroj: http://aboutforensics.co.uk/dna-analysis/ SNOMED CT: 62302004 MeSH: nenalezeno ICD10: nenalezeno
Repetitivn´ı sekvence Definice: Mnohon´asobnˇe opakovan´e sekvence napˇr. nukleotid˚ u v DNA; pˇredstavuj´ı velkou ˇc´ast lidsk´eho genomu. Synonyma: Repetitivn´ı sekvence nukleov´e kyseliny Zdroj: www.lekarske.slovniky.cz SNOMED CT: 51512005 MeSH: D012091 ICD10: nenalezeno
Trombofilie Definice: Vyˇsˇs´ı sklon ke vzniku krevn´ıch sraˇzenin. Zdroj: www.lekarske.slovniky.cz
MeSH: D019851 Tato pr´ace byla podpoˇrena projektem SVV-2015ICD10: D68.5 260158 Univerzity Karlovy v Praze.
Kl´ıˇ cov´ a slova Deoxyribonukleov´ a kyselina
INR Definice: Krevn´ı test pouˇz´ıvan´ y ke stanoven´ı spr´avn´eho d´avkov´an´ı protisr´aˇzliv´ ych l´ek˚ u.
Definice: Druh nukleov´e kyseliny, kter´ a je z´ akladem Synonyma: International Normalized Ratio, standardizovan´ y protrombinov´ y ˇcas dˇediˇcn´e informace. Zdroj: www.urmc.rochester.edu/encyclopedia Synonyma: DNA SNOMED CT: 165581004 Zdroj: www.lekarske.slovniky.cz
MeSH: D019934
SNOMED CT: 24851008
ICD10: nenalezeno
MeSH: D004247
Reference
ICD10: nenalezeno
[1] Geddes L. Fallible DNA evidence can mean prison or freedom. New Sci. 2010; 207 (2773): 8-11
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Stonov´a M. – Nestrukturovan´a data ve zdravotnictv´ı zaloˇzen´em na d˚ ukazech
Nestrukturovan´ a data ve zdravotnictv´ı zaloˇ zen´ em na d˚ ukazech Michaela Stonov´ a1 1
ˇ a republika 1. l´ekaˇrsk´ a fakulta Univerzity Karlovy, Praha, Cesk´
Kontakt: Michaela Stonov´ a 1. l´ ekaˇrsk´ a fakulta Univerzity Karlovy Adresa: Kateˇrinsk´ a 32, 121 08 Praha 2 E–mail:
[email protected]
C´ıle v´ yzkumu
jedineˇcn´ ym ˇc´ıseln´ ym identifik´atorem. Tento krok umoˇznil ochranu soukrom´ı pacient˚ u pˇri souˇcasn´em zachov´an´ı Svat´ ym gr´alem vˇsech l´ekaˇr˚ u a informatik˚ u je syst´em schopnosti prov´azat z´aznamy od jednoho konkr´etn´ıho paysledkem tak byl soubor 386 587 elektronick´ ych pro automatizovan´e stanovov´ an´ı diagn´ oz. Pro jeho vy- cienta. V´ u3 v n´asleduj´ıc´ı semi-strukturovan´e podobˇe: tvoˇren´ı je vˇsak nezbytn´e zvl´ adnout nˇekolik vˇedeck´ ych dis- z´aznam˚ cipl´ın, mezi nˇeˇz patˇr´ı strojov´e zpracov´ an´ı pˇrirozen´eho javˇek;pohlav´ı;pacient;diagn´ oza;4 voln´y text. zyka (NLP).1 V pr˚ ubˇehu jeho v´ yvoje byl vytvoˇren vedlejˇs´ı produkt, kter´ y by mohl b´ yt vyuˇzit pro anal´ yzu velk´ ych 29;Muˇz;675032;H353 ;Pacient byl oˇsetˇren . . . objem˚ u dat (Big Data) [3]. Velikost soubor˚ u se v testovac´ı mnoˇzinˇe pohybovala Zdravotnictv´ı zaloˇzen´e na d˚ ukazech (Evidence-based u (jednoslovn´e soubory s d´elkou Healthcare)2 je kriticky z´ avisl´e na dostateˇcn´em mnoˇzstv´ı v rozmez´ı od nˇekolika byt˚ u (nejvˇetˇs´ı soubory s velikost´ı relevantn´ıch podkladov´ ych dat. L´ekaˇrsk´ a dokumentace za- 26 B) aˇz po nˇekolik kilobyt˚ u se rovnala 727 B a znamenan´a ve formˇe voln´eho textu tvoˇr´ı aˇz 80 % elek- aˇz 15,3 kB). Stˇredn´ı velikost soubor˚ y objem vˇsech 386 587 soubor˚ u ˇcinil 1,6 GB. tronick´e l´ekaˇrsk´e dokumentace (ELD). Jedn´ a se tak o celkov´ L´ekaˇrsk´e z´aznamy byly zpracov´any v syst´emu IBM jedineˇcn´ y zdroj dat, kter´ y byl vˇsak doposud pˇrehl´ıˇzen kv˚ uli sv´e obt´ıˇznˇe analyzovateln´e nestrukturovan´e podobˇe. Watson Content analytics vych´azej´ıc´ıho z projektu Z´ıskan´e znalosti a dovednosti pˇri modelov´ an´ı ˇcesk´eho NLP Apache Lucene. Vˇsechna vstupn´ı data byla standardnˇe umoˇznily vytvoˇrit n´ asleduj´ıc´ı syst´em pro velkoobjemov´e nakrolov´ana,5 rozparsov´ana6 a zaindexov´ana7 [5]. Pro hlubˇs´ı anal´ yzu medic´ınsk´ ych dat bylo vytvoˇreno tˇechto zpracov´an´ı nestrukturovan´e l´ekaˇrsk´e dokumentace. pˇet ad hoc kategori´ı (naz´ yvan´ ych fazety):
Souˇ casn´ y stav pozn´ an´ı Pˇrizp˚ usoben´ı NLP ˇcesk´emu l´ekaˇrsk´emu prostˇred´ı se stalo z´akladn´ım krokem na cestˇe k u ´spˇeˇsn´e anal´ yze nestrukturovan´ ych dat. Pˇri v´ ystavbˇe ˇcesk´eho modelu byl br´an zˇretel nejen na jazykov´e discipl´ıny, jak´ ymi jsou tvaroslov´ı, vˇetn´a skladba a s´emantika, ale rovnˇeˇz na vysokou entropii jazyka l´ekaˇrsk´e dokumentace [7]. Vytvoˇren´ y model byl otestov´ an na elektronick´ ych z´ aznamech pacient˚ u poch´azej´ıc´ıch z nemocniˇcn´ıho informaˇcn´ıho syst´emu ´ redn´ı vojensk´e nemocnice (UVN) ´ (NIS) Ustˇ v Praze. Testovac´ı z´aznamy poch´ azely z r˚ uzn´ ych ambulanc´ı nemocnice (napˇr. pohotovost, kardiologie, koˇzn´ı oddˇelen´ı, oˇcn´ı klinika atd.) Anonymizace z´ aznam˚ u probˇehla jiˇz na u ´rovni extrakce soubor˚ u z NIS. Jm´eno pacienta i jeho oˇsetˇruj´ıc´ıho l´ekaˇre bylo smaz´ ano a rodn´e ˇc´ıslo nahrazeno 1 Natural
Language Processing. jako jeho podmnoˇ zina medic´ına zaloˇ zen´ a na d˚ ukazech (Evidence-based Medicine) [1]. 3 Textov´ e soubory v UTF-8 k´ odov´ an´ı. 4 Dle MKN-10 k´ odu [4]. 2 Stejnˇ e
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
-
Vˇek, Pohlav´ı, Pacient, Diagn´oza, L´eˇciva.
Fazeta Vˇek obsahuje podfazety pro kaˇzdou ˇzivotn´ı dek´adu, dvˇe zvl´aˇstn´ı podfazety pro nezletil´e a zletil´e ´ pacienty do 19 let (d˚ uvodem je skuteˇcnost, ˇze UVN se prim´arnˇe zamˇeˇruje na l´eˇcbu dospˇel´ ych) a kategorii s oznaˇcen´ım Ostatn´ı urˇcenou pro novorozence a pˇr´ıpadn´e chyby v dokumentaci. Fazeta Pohlav´ı rozliˇsuje mezi muˇzsk´ ymi a ˇzensk´ ymi pacienty. Fazeta Pacient umoˇzn ˇuje uk´azat vˇsechny l´ekaˇrsk´e z´aznamy vybran´eho pacienta. Fazeta Diagn´ oza rozdˇeluje nemoci do z´akladn´ıch 22 skupin a jejich 276 podskupin dle MKN-10 k´od˚ u [4]. Posledn´ı 5Z
angl. crawling –naˇ c´ıt´ an´ı dat z vymezen´ eho datov´ eho prostoru. V souˇ casnosti nen´ı ustanoven ofici´ aln´ı pˇreklad tohoto term´ınu. 6 Rozdˇ elen´ı textu na jednotliv´ a slova.V souˇ casnosti nen´ı ustanoven ofici´ aln´ı pˇreklad tohoto term´ınu. 7 Kaˇ zd´ e slovo je zaregistrov´ ano a uloˇ zeny jeho vlastnosti (slovn´ı druh, jazyk, speci´ aln´ı v´ yraz apod.).
47
48
Stonov´a M. – Nestrukturovan´a data ve zdravotnictv´ı zaloˇzen´em na d˚ ukazech
fazeta L´eˇciva pˇredstavuje ATC (Anatomical Therapeutic Chemical) klasifikaˇcn´ı syst´em l´eˇciv, kter´ yu ´ˇcinn´e l´atky zaˇcleˇ nuje do r˚ uzn´ ych skupin na z´ akladˇe jejich org´ anov´eho a syst´emov´eho p˚ usoben´ı a dle jejich terapeutick´ ych, farmaceutick´ ych a chemick´ ych vlastnost´ı. Celkem 5 862 podfazet kop´ıruje rozdˇelen´ı l´eˇciv do 14 z´ akladn´ıch skupin (prvn´ı u ´roveˇ n) a jejich pˇr´ısluˇsn´ ych podskupin aˇz do p´at´e moˇzn´e u ´rovnˇe. Na z´akladˇe posledn´ıch dvou fazet je syst´em v souˇcasn´em stavu schopen rozpoznat a zaˇradit 38 617 diagn´oz dle MKN-10 kod˚ u [4] a 4 2568 l´eˇciv registrovan´ ych St´atn´ım u ´stavem pro kontrolu l´eˇciv, vˇcetnˇe napˇr. nejnovˇeji pˇridan´ ych poloˇzek v ATC skupinˇe L01X jin´ych cytostatik.9 Tabulka 1: Pacienti kuˇra ´ci.
V pr˚ ubˇehu zpracov´ an´ı bylo kaˇzd´e slovo jak lingvisticky rozpozn´ano, tak i pˇr´ısluˇsnˇe oznaˇceno, v pˇr´ıpadˇe, ˇze bylo zaˇrazeno do nˇekter´e z fazet ˇci podfazet. Vˇsechny tyto vedlejˇs´ı informace byly uloˇzeny v 69 souborech indexu o celkov´e velikosti 4,8 GB. Pˇet pˇridan´ ych fazet a standardn´ı NLP zpracov´an´ı nav´ yˇsilo index do t´e m´ıry, ˇze se p˚ uvodn´ı velikost vstupn´ıch dat (1,6 GB) ztrojn´ asobila.
Poˇcet pacient˚ u pˇrich´azej´ıc´ıch na ambulanci v podnapil´em stavu m˚ uˇze b´ yt dalˇs´ı u ´lohou pro anal´ yzu nestrukturovan´eho textu. Pouze 244 l´ekaˇrsk´ ych z´aznam˚ u bylo MKN-10 k´odem jasnˇe oznaˇceno jako pˇr´ımo souvisej´ıc´ı s poˇz´ıv´an´ı alkoholu (MKN-10 k´ody F100 a F102), avˇsak dalˇs´ıch 549 z´aznam˚ u v textu v´ yslovnˇe zmiˇ novalo, ˇze byl y. Celkovˇe tedy alespoˇ n 793 patient˚ u nebylo Jednou jiˇz zaindexovan´ a data je moˇzno ihned analyzo- pacient opil´ vat v IBM Watson Content Analytics grafick´em rozhran´ı. v dobˇe n´avˇstˇevy ambulance ve stˇr´ızliv´em stavu (Tabulka 12 Pouh´ ym rozkliknut´ım pˇr´ısluˇsn´e fazety lze napˇr. zjistit, 2). ˇze z 386 587 z´aznam˚ u, 211 586 pacient˚ u bylo muˇzsk´eho pohlav´ı a 175 001 ˇzensk´eho. Vˇetˇsina z pacient˚ u (58 026 pacient˚ u) navˇst´ıvila ambulance opakovanˇe, zbyl´ı zde byli oˇsetˇreni pouze jednou (35 898 pacient˚ u). Celkovˇe proˇslo ´ ambulancemi UVN 93 924 r˚ uzn´ ych pacient˚ u, kteˇr´ı byli l´eˇceni na z´akladˇe 4 525 rozd´ıln´ ych diagn´ oz.10 Pˇredchoz´ıch v´ ysledk˚ u bylo moˇzno dos´ ahnout i cestou standardn´ıch datab´ azov´ ych dotaz˚ u, nebot’ prvn´ı ˇctyˇri ˇc´asti elektronick´ ych z´aznam˚ u jsou ve strukturovan´e formˇe. Zjiˇstˇen´ı, kolik kuˇr´ak˚ u mezi pacienty bylo l´eˇceno jin´ymi cytostatiky (L01X ATC skupina) a pro jakou diagn´ozu, jiˇz nen´ı t´ımto pˇr´ıpadem. Informace o tom, zda je ˇci nen´ı pacient kuˇr´akem, je uloˇzena ve voln´em textu l´ekaˇrsk´e dokumentace stejnˇe jako pˇredepsan´e l´eky. Anal´ yza na z´akladˇe obsahu textu nestrukturovan´ ych dat je tak jedinou moˇznost´ı, jak dos´ ahnout v´ ysledku. Z´ akladn´ı model ˇcesk´eho NLP dok´aˇze filtrovat slova nejen na z´ akladˇe jejich z´akladn´ıch tvar˚ u, jimiˇz jsou v tomto pˇr´ıpadˇe kuˇr´ ak nebo kuˇraˇcka nebo kouˇrit, ale automaticky rovnˇeˇz i dle jejich odvozen´ ych forem.11 Anal´ yza na z´ akladˇe NLP objevila, ˇze l´ekaˇrsk´e z´aznamy v´ yslovnˇe zmiˇ nuj´ı 13 484 kuˇr´ ak˚ u, z nichˇz 15 bylo pod´ano l´eˇcivo z ATC skupiny L01X. V´ ysledky, vˇcetnˇe pˇr´ısluˇsn´ ych diagn´ oz, jsou uvedeny v Tabulce 1.
8 St´ atn´ı
u ´stav pro kontrolu l´ eˇ civ m´ a zaregistrov´ ano t´ emˇ eˇr 56 000 l´ eˇ civ. Vˇ etˇsina z nich je vˇsak pod stejn´ ym jm´ enem vyr´ abˇ ena v´ıce farmaceutick´ ymi firmami. Tyto duplicity umoˇ znily sn´ıˇ zit p˚ uvodn´ı poˇ cet na hodnotu 4 256. 9 Zejm´ ena se jednalo o monoklon´ aln´ı protil´ atky a r˚ uzn´ e inhibitory. 10 Dle MKN-10 k´ od˚ u. 11 Napˇ r. kouˇ r´ı, kouˇ ril, kouˇ rila, kuˇ r´ ack´ a, kuˇ r´ akem, kuˇ r´ ack´ y atd.
Tabulka 2: Pacienti v podnapil´em stavu.
Navrˇzen´ y syst´em umoˇzn ˇuje rychlou anal´ yzu13 nestrukturovan´ ych dat. Syst´em je schopn´ y pracovat v reˇzimu Big Data, a to aˇz do velikosti v jednotk´ach terabyt˚ u. Pot´e, co jsou data jiˇz jednou nakrolov´ana a zaindexov´ana, anal´ yza m˚ uˇze b´ yt provedena v re´aln´em ˇcase. Jedin´ ym omezen´ım je vysok´a entropie jazyka l´ekaˇrsk´e dokumentace. V nˇekter´ ych pˇr´ıpadech m˚ uˇze tato nejasnost pozmˇenit v´ ysledky anal´ yzy. Tento jev a jeho dopad je pˇr´ımo vidˇet v obou dvou tabulk´ach. Exponenci´aln´ı z´apis laboratorn´ıch v´ ysledk˚ u ve formˇe x10 (napˇr. Trombocyty: 12 Tabulka
ˇ c. 2 ukazuje pouze diagn´ ozy, jejichˇ zˇ cetnost byla vˇ etˇs´ı nebo rovna hodnotˇ e 10. 13 K proveden´ ı anal´ yzy kouˇr´ıc´ıch pacient˚ u i pacient˚ u v podnapil´ em stavu bylo zapotˇreb´ı dvou minut, a to vˇ cetnˇ e exportu v´ ysledn´ ych dat do csv souboru. 14 Syst´ em z´ amˇ ernˇ e nerozliˇsuje velk´ a a mal´ a p´ısmena.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Stonov´a M. – Nestrukturovan´a data ve zdravotnictv´ı zaloˇzen´em na d˚ ukazech
176.00 x10ˆ9/l) je totoˇzn´ y14 se z´ apisem MKN-10 k´odu X10 pro Kontakt s hork´ymi n´ apoji, potravou, tukem a oleji na vaˇren´ı. Vysok´ y poˇcet X10 diagn´ oz je v obou dvou pˇr´ıpadech chybn´ y, ale v celkov´em poˇctu soubor˚ u zanedbateln´ y. Tento typ chyby je v´ıce pravdˇepodobn´ yu kr´atk´ ych zkratek, avˇsak i delˇs´ı slova, jak´ ym je napˇr. ketokonazole mohou zp˚ usobit jist´e dezinterpretace. Ketokonazole m˚ uˇze b´ yt jak topick´ ym antimykotikem (ATC skupina D01AC08), tak i hormon´ aln´ım supresantem (ATC skupina H02C). O jak´ y pˇr´ıpad se jedn´ a, lze v obou pˇr´ıpadech zjistit pouze z kontextu. Navzdory skuteˇcnosti, ˇze zm´ınˇen´e chyby maj´ı mal´ y dopad na v´ ysledky v rozsahu Big Data, pˇr´ıˇst´ı kroky v´ yzkumu povedou ke zm´ırnˇen´ı jejich dopadu. Dalˇs´ı v´ yzkum tak bude zamˇeˇren na zlepˇsen´ı kontextov´e anal´ yzy a opravu pravopisn´ ych chyb.
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı
49
Synonyma: Evidence-based Medicine Zdroj: Cochrane collaboration http://www.cochrane. org/ SNOMED CT: nenalezeno MeSH: D057286 ICD10: nenalezeno
NLP Definice: Odvˇetv´ı informatiky, kter´e se zab´ yv´a automatizovan´ ym zpracov´an´ım pˇrirozen´eho jazyka. Synonyma: Human Language Technology, Natural Language Processing
Zdroj: The Free Dictionary http://www.thefreedictionary. com/natural+language+processing Navrˇzen´ y syst´em je schopen zpracovat terabyty l´ekaˇrsk´e dokumentace a je nez´ avisl´ y na form´ atu vstupn´ıch dat. Lze jej tak vyuˇz´ıt bez nutnosti dalˇs´ıho pˇrizp˚ usoben´ı. SNOMED CT: nenalezeno Automatick´a klasifikace nestrukturovan´ ych dat na z´akladˇe vˇeku, pohlav´ı, diagn´ ozy a pouˇzit´ ych l´eˇciv je jiˇz MeSH: D009323 souˇc´ast´ı z´akladn´ıho nastaven´ı. Syst´em tak umoˇzn ˇuje ICD10: nenalezeno prov´adˇet pˇrehledovou anal´ yzu l´ekaˇrsk´e dokumentace a dalˇs´ıch nestrukturovan´ ych zdravotnic´ıch dat s dostaˇcuj´ıc´ı a data pˇresnost´ı. Zm´ınˇen´e nepˇresnosti zp˚ usoben´e vyˇsˇs´ı entropi´ı Nestrukturovan´ jazyka l´ekaˇrsk´e dokumentace budou pˇredmˇetem dalˇs´ıho Definice: Data, kter´a nejsou uchov´av´ana v datab´azi v´ yzkumu. nebo v jinak pˇreddefinovan´e struktuˇre.
Podˇ ekov´ an´ı
Synonyma: Nestrukturovan´e informace
Tato pr´ace byla podpoˇrena projektem SVV-2015- Zdroj: The Free Dictionary http://encyclopedia. 260158 Univerzity Karlovy v Praze. thefreedictionary.com/Unstructured+data
Kl´ıˇ cov´ a slova Big Data
SNOMED CT: nenalezeno MeSH: nenalezeno
ICD10: nenalezeno Definice: Velkobjemov´e zpracov´ an´ı rozd´ıln´ ych druh˚ u dat v re´aln´em (ˇci pˇrimˇeˇrenˇe akceptovateln´em) ˇcase. Strukturovan´ a data Zdroj: Gartner http://www.gartner.com/it-glossary/ Definice: Data, kter´a jsou uchov´avan´a v pˇresnˇe definobig-data van´e struktuˇre. SNOMED CT: nenalezeno Synonyma: Strukturovan´e informace MeSH: nenalezeno Zdroj: Webopedia http://www.webopedia.com/TERM/ ICD10: nenalezeno S/structured_data.html
Medic´ına zaloˇ zen´ a na d˚ ukazech
SNOMED CT: nenalezeno
Definice: Vˇedom´e, zˇreteln´e a soudn´e pouˇz´ıv´an´ı nej- MeSH: nenalezeno lepˇs´ıch souˇcasn´ ych d˚ ukaz˚ u pˇri rozhodov´an´ı o p´eˇci o jednotliv´e pacienty. ICD10: nenalezeno S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
50
Stonov´a M. – Nestrukturovan´a data ve zdravotnictv´ı zaloˇzen´em na d˚ ukazech
Reference [1] Cochrane Collaboration, Available from: https://www.cochrane.org. [2] Hartzband P, Groopman J. Untangling the Web –Patients, Doctors, and the Internet. N Engl J Med 2010; 362:1063–1066. [3] Holzinger A, Stocker C, Ofner B, Prochaska G, Brabenetz A, Hofmann-Wellenhof R. Combining HCI, Natural Language Processing, and Knowledge Discovery –Potential of IBM Content Analytics as an Assistive Technology in the Biomedical Field. Human-Computer Interaction and Knowledge Discovery in Complex, Unstructured, Big Data.2013; 7947: 13–24.
[4] Institute of Health Information and Statistics of the Czech Republic, Available from: https://www.uzis.cz. [5] Stonov´ a M. Unstructured Data in Healthcare. Semantic Interoperability in Biomedicine and Healthcare. IJBH 2014; 2(1): 34–36. [6] State Institute for https://www.sukl.eu.
Drug
Control,
Available
from:
[7] Zvolsk´ y M. Automating the Use of Clinical Practice Guidelines in the Health Information Infrastructure. Semantic Interoperability in Biomedicine and Healthcare. IJBH 2014; 2(1): 51–52.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
51
ˇ nkov´a B. a kol. – V´yvoj pˇrevodn´ıho syst´emu u myˇsi Saˇ
V´ yvoj pˇrevodn´ıho syst´ emu u myˇsi ˇ nkov´ Barbora Saˇ a1,2 , Jiˇr´ı Beneˇs1,2 , David Sedmera1,2 1
ˇ a Republika Anatomick´y u ´stav, 1. l´ekaˇrsk´ a fakulta, Univerzita Karlova v Praze, Praha, Cesk´ 2
ˇ v.v.i., Praha, Cesk´ ˇ a Republika Fyziologick´y u ´stav, AVCR,
Kontakt: ˇ nkov´ Barbora Saˇ a Anatomick´ yu ´stav, 1. l´ ekaˇrsk´ a fakulta Adresa: U Nemocnice 3, 128 08 Praha 2 E–mail:
[email protected]
C´ıle v´ yzkumu Pˇrevodn´ı syst´ em srdeˇ cn´ı – v´ yvoj a funkce Elektrick´ y impulz se tvoˇr´ı ve specializovan´e tk´ani pacemakeru, sinoatrialn´ım uzlu, um´ıstˇen´em v prav´e pˇreds´ıni u vstupu horn´ı dut´e ˇz´ıly. Z m´ısta vzniku postupuje elektrick´ y vzruch rychle vedouc´ı tk´ an´ı myokardu pˇreds´ın´ı aˇz do atrioventrikul´ arn´ıho uzlu, um´ıstˇen´eho na rozhran´ı pˇreds´ın´ı a komor. D´ıky pomal´emu veden´ı vzruchu atrioventrikul´arn´ım uzlem doch´ az´ı ke vzniku zpoˇzdˇen´ı, nezbytn´eho pro efektivn´ı plnˇen´ı komor bˇehem diastoly. Vzruch je ˇs´ıˇren ze sinoatrialn´ıho uzlu rychle vedouc´ımi ˇc´astmi pˇrevodn´ıho syst´emu srdeˇcn´ıho Hisov´ ym svazkem, n´aslednˇe Tawarov´ ymi ram´enky aˇz do dist´ aln´ıch vˇetviˇcek Purkyˇ nov´ ych vl´aken, kter´e aktivuj´ı pracovn´ı myokard komor. Silnˇejˇs´ı svazky komorov´eho pˇrevodn´ıho syst´emu jsou elektricky izolovan´e od okoln´ı tk´ anˇe. K naruˇsen´ı izolace doch´az´ı v dist´ aln´ıch ˇc´ astech Purkyˇ nova syt´emu, to umoˇzn ˇuje pˇr´ım´e spojen´ı s pracovn´ım myokardem s n´aslednou aktivac´ı myokardu komor smˇerem od apexu k b´azi a tak´e aktivace od endokardu k epikardu. Jednotliv´e ˇc´asti srdeˇcn´ıho pˇrevodn´ıho syst´emu se bˇehem v´ yvoje diferencuj´ı specifick´ ym zp˚ usobem, kter´ y je morfogeneticky zachov´ an. Tato morfogenetick´a posloupnost se shoduje u kuˇrete, myˇsi i u ˇclovˇeka [4]. Pacemaker je prvn´ı funkˇcn´ı odd´ıl pˇrevodn´ıho syst´emu identifikovateln´ y u st´adia trubicovit´eho srdce, u myˇsi k tomu doch´az´ı kolem embryon´ aln´ıho dne (ED) 7,5. Srdeˇcn´ı impulz se ˇs´ıˇr´ı pomalu, izotropn´ım zp˚ usobem ze sinus venosus, um´ıstˇen´eho kaud´ alnˇe, primitivn´ı komorou d´ale krani´aln´ım smˇerem k v´ ytokov´e ˇc´ asti [5]. Ve st´ adiu srdeˇcn´ı kliˇcky (myˇs ED 8) se objev´ı pomalu vodiv´ y atrioventrikul´arn´ı kan´al, kter´ y oddˇeluje rychleji vodiv´e oblasti pˇreds´ın´ı a komor. Segmenty pomalu vedouc´ı elektrick´ y vzruch (atrioventrikul´ arn´ı kan´ al, v´ ytokov´ a ˇca´st a sinoatri´aln´ı oblast) funguj´ı jako svˇeraˇce, jejichˇz kontrakce jsou koordinovan´e, a t´ım je zv´ yˇsena efektivita krevn´ı cirkulace. Komorov´a aktivaˇcn´ı sekvence ve f´ azi srdeˇcn´ı kliˇcky kop´ıruje pr˚ utok krve. Ve v´ yvojov´em obdob´ı pˇred septac´ı srdce (myˇs ED 9,5–12,5) se objevuje primitivn´ı aktivaˇcn´ı S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
vzor prim´arn´ı interventrikul´arn´ı prstenec“ (primary in” terventricular ring), pˇrednostnˇe vedouc´ı aktivaci pod´el budouc´ı mezikomorov´e pˇrep´aˇzky [11]. Rychle vodiv´ y komorov´ y pˇrevodn´ı syst´em je posledn´ı ˇc´ast, kter´a se diferencuje, jej´ıˇz funkˇcnost se projev´ı zmˇenou ventrikul´arn´ı aktivaˇcn´ı sekvence. Nezral´a aktivaˇcn´ı sekvence kop´ıruj´ıc´ı tok krve od srdeˇcn´ı b´aze smˇerem k apexu je nahrazena aktivac´ı opaˇcn´ ym smˇerem od vrcholu k b´azi. Zral´a aktivaˇcn´ı sekvence od apexu k b´azi se objevuje s dokonˇcen´ ym interventrikul´arn´ım septem u kuˇrete [3, 8], zat´ımco u myˇsi se mˇen´ı aktivace dˇr´ıve ve v´ yvoji, pˇred dokonˇcen´ım interventrikul´arn´ıho septa [9]. Funkˇcnost srdeˇcn´ıho pˇrevodn´ıho syst´emu lze nejl´epe hodnotit pomoc´ı vysokorychlostn´ıho optick´eho mapov´an´ı, hlavn´ı metody pouˇzit´e v t´eto studii. Bˇehem optick´eho mapov´an´ı je pouˇzit syst´em virtu´aln´ıch elektrod, fluorescenˇcn´ı barvivo rychle reaguj´ıc´ı na zmˇeny elektrick´eho potenci´alu a vysokorychlostn´ı kamera slouˇz´ıc´ı ke sbˇeru dat. Optick´e mapov´an´ı poskytuje in vivo fyziologick´e u ´daje – epikardi´aln´ı aktivaˇcn´ı mapy, vznikaj´ıc´ı anal´ yzou dat z optick´eho mapov´an´ı a berouc´ı v u ´vahu ˇcasoprostorovou osu. Z aktivaˇcn´ı mapy mohou b´ yt vyhodnoceny prvn´ı aktivovan´a m´ısta na epikardi´aln´ım povrchu a smˇer veden´ı elektrick´eho vzruchu, dohromady tvoˇr´ıc´ı vzor komorov´e aktivaˇcn´ı sekvence, ukazatel zralosti pˇrevodn´ıho syst´emu. Dalˇs´ım parametrem, kter´ y lze hodnotit z aktivaˇcn´ıch map, je rychlost ˇs´ıˇren´ı elektrick´eho impulzu epikardem.
Norm´ aln´ı v´ yvoj myˇs´ıho pˇrevodn´ıho syst´ emu Morfologie a v´ yvoj pˇrevodn´ıho syst´emu u myˇsi byly pops´any histologicky v sedmdes´at´ ych letech, nejnovˇeji vˇsak optick´e mapov´an´ı umoˇznuje studium jeho funkˇcn´ıho zapojen´ı. V souˇcasn´e dobˇe chyb´ı kvantitativn´ı hodnocen´ı aktivaˇcn´ıch ˇcas˚ u komor a aktivaˇcn´ıch vzor˚ u, nezbytn´ ych pro interpretaci zmˇen pozorovateln´ ych u transgenn´ıch myˇs´ıch model˚ u, kde lze pˇredpokl´adat moˇzn´ y vliv na v´ yvoj pˇrevodn´ıho syst´emu. Pomoc´ı metody optick´eho mapov´an´ı jsme studovali funkci pˇrevodn´ıho syst´emu u myˇsi poˇc´ınaje ED 9,5. Mˇeˇrili jsme celkov´ y aktivaˇcn´ı ˇcas lev´e komory a hodnotili zp˚ usob
52
ˇ nkov´a B. a kol. – V´yvoj pˇrevodn´ıho syst´emu u myˇsi Saˇ
aktivace komor zaˇrazen´ım do jedn´e kategorie aktivaˇcn´ıch vzor˚ u; aktivace vyuˇz´ıvaj´ıc´ı prim´ arn´ı interventrikul´arn´ı prstenec, lev´eho Tawarova ram´enka (LBB), prav´eho Tawarova ram´enka (RBB), obou ram´enek, pˇrechodn´ y zp˚ usob. Typick´a situace v dan´em ED je reprezentov´ana spektrem nˇekolika aktivaˇcn´ıch vzor˚ u, a proto nen´ı moˇzn´e popsat norm´aln´ı v´ yvoj anal´ yzou m´enˇe neˇz 10 embry´ı ve skupinˇe. Prvn´ı primitivn´ı aktivaˇcn´ı vzor, kter´ y se objev´ı v ˇcasn´em v´ yvoji pˇrevodn´ıho syt´emu u myˇsi, je aktivace pomoc´ı prim´arn´ıho interventrikul´ arn´ıho prstence. Na rozd´ıl od kuˇrete, kde je prvn´ı aktivace od b´ aze k apexu, nebyl typ aktivace od b´aze k apexu nikdy pozorov´ an u myˇsi. Veden´ı elektrick´eho impulzu v rozmez´ı ED 9–11 je charakteristick´e t´ım, ˇze je vyuˇzit prim´ arn´ı interventrikul´ arn´ı prstenec, struktura v oblasti budouc´ıho interventrikul´ arn´ıho septa. Jak pokraˇcuje v´ yvoj srdce, postupnˇe miz´ı aktivace pomoc´ı prim´arn´ıho interventrikul´ arn´ıho prstence (ED 12,5 se vyskytuje naposled) a je nahrazov´ ana aktivac´ı pomoc´ı Tawarov´ ych ram´enek. Z komorov´eho pˇrevodn´ıho syst´emu je prvn´ı aktivn´ı LBB, ale s v´ yskytem pˇrechodn´eho typu aktivace. V ED 11,5 se st´ av´ a aktivn´ı RBB nebo obˇe ram´enka. V ED 14,5 je vˇetˇsina srdc´ı aktivovan´ ych ze dvou center, ale ojedinˇele bylo prvn´ı aktivovan´e m´ısto jen na ˇıˇren´ı elektrick´eho impulzu se urychluje jedn´e stranˇe. S´ nejv´ yraznˇeji ve v´ yvojov´em oknˇe mezi ED 10,5–12,5, coˇz vypl´ yv´a z porovn´an´ı aktivaˇcn´ıch ˇcas˚ u u jednotliv´ ych ED. Aktivaˇcn´ı ˇcas z˚ ust´av´a v pozdˇejˇs´ıch st´ adi´ıch na stejn´ ych hodnot´ach jako v pˇredchoz´ıch ED, ale nen´ı zohlednˇen fakt, ˇze srdce st´ale roste. Zhodnocen´ı norm´aln´ıho v´ yvoje aktivace komor poskytuje nezbytn´ y r´amec pro anal´ yzu transgenn´ıch myˇs´ı s potenci´aln´ımi vadami pˇrevodn´ıho syst´emu. Pro interpretaci zmˇen pozorovan´ ych u myˇs´ıch model˚ u je nutn´e systematick´e a kvantitativn´ı studium aktivaˇcn´ıch sekvenc´ı. Jako pozad´ı byla pouˇzita z´ıskan´ a data bˇehem anal´ yz myˇs´ı postr´adaj´ıc´ı geny pro Cx40 [11, 2], Tbx2, Ptx2 [1], ErbB2 a v myˇs´ım modelu se syndromem dlouh´eho QT intervalu [10].
Nedostatek Cx40 Cx40 je hlavn´ı protein vodiv´ ych bunˇeˇcn´ ych spoj˚ u (gap junction) exprimovan´ y v pˇreds´ın´ıch a komor´ ach v ˇcasn´ ych f´az´ıch embryogeneze. Jak embryogeneze pokraˇcuje, exprese Cx40 v pˇreds´ın´ıch se nemˇen´ı, ale v komor´ ach se postupnˇe omezuje na trabekuly, aˇz je dosaˇzeno exprese pouze v komorov´em pˇrevodn´ım syst´emu u dospˇel´ ych (Tawarova ram´enka, Purkyˇ nova vl´ akna). Nedostatek Cx40 zp˚ usobuje embryon´aln´ı a atri´aln´ı anom´ alie ve veden´ı elektrick´eho impulzu [7], pomalejˇs´ı veden´ı vzruchu a blok prav´eho Tawarova ram´enka [13, 15, 6]. Nicm´enˇe funkˇcn´ı v´ yznam Cx40 bˇehem v´ yvoje pˇrevodn´ıho syst´emu nebyl doposud studov´an. Ve ED 12,5 a ED 14,5 bylo pozorov´ ano r˚ uznorod´e veden´ı rychlosti vzruchu pˇreds´ınˇemi v z´ avislosti na Cx40 genotypu. U divok´eho typu (WT) a heterozygotn´ıch embry´ı bylo prvn´ı aktivovan´e m´ısto v oblasti sinoatri´ aln´ıho uzlu, u heterozygot˚ u byl m´ırnˇe prodlouˇzen´ y ˇcas epikardi´aln´ı
aktivace (statistick´ y v´ yznamn´ y ED 12,5). U myˇs´ı bez Cx40 se objevily dvˇe atri´aln´ı aktivaˇcn´ı sekvence, aktivace ze sinoatri´aln´ıho uzlu a ektopick´a aktivace poch´azej´ıc´ı z prav´eho ouˇska. D´ale se vliv nedostatku Cx40 projevil prodlouˇzen´ım doby aktivace, kter´a byla v´ yraznˇe zpomalena hlavnˇe u ektopicky aktivovan´ ych pˇreds´ın´ı a u ED 12,5 v porovn´an´ı s ED 14,5. Chybˇej´ıc´ı Cx40 bˇehem v´ yvoje ovlivˇ nuje aktivaˇcn´ı sekvence ˇs´ıˇren´ı vzruchu v atri´ıch a v´ yraznˇe zpomaluje rychlost ˇs´ıˇren´ı, kter´a je v pˇr´ım´e korelaci s typem atri´aln´ı aktivace. Rozd´ıly mezi genotypy pˇri anal´ yze aktivaˇcn´ıch ˇcas˚ u komor nebyly pozorov´any. Sledov´an´ı ˇcetnosti aktivaˇcn´ıch sekvenc´ı u komor odhalilo znaˇcn´ y pokles frekvence aktivace pomoc´ı lev´eho Tawarova ram´enka na ED 12,5 a ED 14,5 u Cx40 deficientn´ıch myˇs´ı. U heterozygot˚ u byla ˇcetnost v´ yskytu tak´e sn´ıˇzena, ale v menˇs´ım rozsahu. Pod´ıl aktivaˇcn´ıch sekvenc´ı se zaˇcal obracet v ED 16,5, kde byla aktivn´ı LBB pˇr´ıtomna u vˇsech genotyp˚ u v t´emˇeˇr pln´em rozsahu. Frekvence pravostrann´e aktivace RBB zaˇcala klesat u myˇs´ı s Cx40 deficitem, coˇz naznaˇcuje rozvoj dysfunkce prav´eho Tawarova ram´enka. Nejv´ yraznˇejˇs´ı fenotyp byl zaznamen´an v ED 18,5, kde srdce bez Cx40 mˇela funkˇcn´ı RBB pouze v 33% pˇr´ıpadech na rozd´ıl od 96% u heterozygot˚ u a 94% WT.
Souˇ casn´ y stav pozn´ an´ı Studie byla dokonˇcena a data publikov´ana [11, 1, 10, 2]. Optick´e mapov´an´ı a whole mount konfok´aln´ı mikroskopie pˇredstavuj´ı nejmodernˇejˇs´ı techniky pro studium fyziologick´ ych a morfologick´ ych vlastnost´ı embryon´aln´ıch tk´an´ı. Obˇe tyto metody jsou nyn´ı dobˇre zaveden´e na 1.LF UK.
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı Pochopen´ı mechanism˚ u signalizace v´ yvoje pˇrevodn´ıho syst´emu m˚ uˇze m´ıt v´ yznam pro klinick´e pracovn´ıky a vˇedce v z´akladn´ım v´ yzkumu studuj´ıc´ı srdeˇcn´ı onemocnˇen´ı u dospˇel´ ych. Vrozen´e vady spolu s ektopickou nebo nevhodnou indukc´ı pˇrevodn´ıch tk´an´ı jsou procesy, kter´e pˇrisp´ıvaj´ı k arytmi´ı u dospˇel´ ych. Arytmie pˇredstavuj´ı skupinu kardiovaskul´arn´ıch chorob, kter´e v´ yznamnˇe pˇrisp´ıvaj´ı k morbiditˇe a mortalitˇe v populaci.
Podˇ ekov´ an´ı Tato studie byla ˇc´asteˇcnˇe finanˇcnˇe podpoˇrena z prostˇredk˚ u SVV-2015-260158 projekt Univerzity Karlovy, MSMT VZ 0021620806, AS CR AVOZ50450515, GACR 304/08/0615. S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
53
ˇ nkov´a B. a kol. – V´yvoj pˇrevodn´ıho syst´emu u myˇsi Saˇ
Kl´ıˇ cov´ a slova Optick´ e mapov´ an´ı Definice: Zobrazovan´ı na napˇet´ı z´ avisl´ ym barvivem. Synonyma: Mapov´ an´ı akˇcn´ıho potenci´ alu Zdroj: [12] SNOMED CT: nenalezeno MeSH: D056969 ICD10: nenalezeno
Blok prav´ eho Tawarova ram´ enka Definice: Forma srdeˇcn´ıho bloku, kdy je elektrick´ y impulz pro komory pˇreruˇsen. Synonyma: Blok ram´enka Zdroj: [14] SNOMED CT: 404684003 MeSH: D002037 ICD10: I45.1
Konexin 40 Definice: Protein spojen´ı typu gap junction“. ” Synonyma: transmembranov´ y protein genu Gja5 Zdroj: [6] SNOMED CT: nenalezeno MeSH: C082919 ICD10: nenalezeno
Reference [1] Ammirabile G, Tessari A, Pignataro V, Szumska D, Sardo FS, Benes J, Jr., Balistreri M, Bhattacharya S, Sedmera D, Campione M. 2012. Pitx2 confers left morphological, molecular, and functional identity to the sinus venosus myocardium. Cardiovasc Res 93:291-301.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
[2] Benes J Jr, Ammirabile G, Sankova B, Campione M, Krejci E, Kvasilova A, Sedmera D. 2014. The role of connexin40 in developing atrial conduction. FEBS Lett. 2014 Apr 17;588(8):14659. [3] Chuck ET, Freeman DM, Watanabe M, Rosenbaum DS. 1997. Changing activation sequence in the embryonic chick heart. Implication for the development of the His-Purkinje system. Circ Res 81:470-476. [4] Gourdie RG, Harris BS, Bond J, Justus CHKW, O’Brien TX, Thompson RP, Sedmera D. 2003. Development of the Cardiac Pacemaking and Conduction System. Birth Defects Research 69:46-57. [5] Kamino K. 1991. Optical approaches to ontogeny of electrical activity and related functional organization during early heart development. Physiol Rev 71:53-91. [6] Kirchhoff S, Nelles E, Hagendorff A, Kruger O, Traub O, Willecke K. Reduced cardiac conduction velocity and predisposition to arrhythmias in connexin40-deficient mice. Curr Biol 1998;8:299–302. [7] Leaf DE, Feig JE, Vasquez C, Riva PL, Yu C, Lader JM et al. Connexin40 imparts conductionheterogeneity to atrial tissue. Circ Res 2008;103:1001–1008. [8] Reckova M, Rosengarten C, deAlmeida A, Stanley CP, Wessels A, Gourdie RG, Thompson RP, Sedmera D. 2003. Hemodynamic is a key epigenetic factor in development of the cardiac conduction system. Circ Res 93:77-85. [9] Rentschler S, Vaidya DM, Tamaddon H, Degenhardt K, Sassoon D, Morley GE, Janife J, Fishmann GI. 2001. Visualization and functional characterization of the developing murine cardiac conduction system. Development 128:1785-1792. [10] de la Rosa AJ, Dominguez JN, Sedmera D, Sankova B, HoveMadsen L, Franco D, Aranega A. 2013. Functional suppression of Kcnq1 leads to early sodium channel remodeling and cardiac conduction system dysmorphogenesis. Cardiovasc Res. 2013 Jun 1;98(3):504-14. [11] Sankova B, Benes JJ, Krejci E, Dupays L, Thevenian-Ruissy M, Miquerol L, Sedmera D. 2012. The effect of connexin40 deficiency on ventricular conduction system function during development. Cardivas Res:doi:10.1093/cvr/cvs1210. [12] Sedmera D, Reckova M, Rosengarten C, Torres MI, Gourdie RG, Thompson R P. 2005. Optical Mapping of Electrical Activation in the Developing Heart. Micros Microanal 11: 209-215. [13] Simon AM, Goodenough DA, Paul DL. Mice lacking connexin40 have cardiac conduction abnormalities characteristic of atrioventricular block and bundle branch block. Curr Biol 1998;8:295–298. [14] Stejfa M et al. 2006. Kardiologie. Grada. 776p. [15] Tamaddon HS, Vaidya D, Simon AM, Paul DL, Jalife J, Morley GE. High-resolution optical mapping of the right bundle branch in connexin40 knockout mice reveals slow conduction in the specialized conduction system. Circ Res 2000;87:929–936.
54
ˇ Steffl M., Plz´ak J. – V´yskyt mikrobi´aln´ı fl´ory nosn´ıch pr˚ uchod˚ u a jej´ı vliv na rozvoj chronick´ych rhinosinusitid
V´ yskyt mikrobi´ aln´ı fl´ ory nosn´ıch pr˚ uchod˚ u a jej´ı vliv na rozvoj chronick´ ych rhinosinusitid 1 ˇ Martin Steffl , Jan Plz´ ak1 1
Klinika otorinolaryngologie a chirurgie hlavy a krku, 1. LF UK, FN v Motole
Kontakt: ˇ Martin Steffl Klinika otorinolaryngologie a chirurgie hlavy a krku, 1. LF UK, FN v Motole ´ Adresa: V Uvalu 84, 150 06 Praha 5 E–mail:
[email protected]
C´ıle v´ yzkumu
do dvou fenotyp˚ u, a to v z´avislosti na pˇr´ıtomnosti nosn´ıch polyp˚ u. Konkr´etnˇe tedy na chronickou rhinosinusitidu u a chronickou rinosinusitidu bez C´ılem projektu je anal´ yza mikrobi´ aln´ıch komunit, vy- s pˇr´ıtomnost´ı polyp˚ u [3]. skytuj´ıc´ıch se v dutin´ach nosn´ıch a jejich porovn´ an´ı u pa- pˇr´ıtomnosti polyp˚ cient˚ u s chronickou rhinosinusitidou at’ jiˇz s polypy, ˇci bez nich, oproti zdrav´ ym kontrol´ am, ˇcili pacient˚ um bez postiˇzen´ı sinonas´aln´ıho traktu. Vˇedeckou ot´ azkou tedy Jedn´a se o onemocnˇen´ı v´ yraznˇe zhorˇsuj´ıc´ı kvalitu jest: Jakou roli hraj´ı zmˇeny ve sloˇzen´ı mikrofl´ ory du” tiny nosn´ı pˇri vzniku chronick´ ych rhinosinusitid, jejich ˇzivota a sniˇzuj´ıc´ı pracovn´ı produktivitu. Jsou spojen´e uchodnost´ı, bolestmi hlavy, podtyp˚ u a jak´a je jejich pˇr´ıpadn´ a korelace s klinick´ ym zejm´ena se zhorˇsenou nosn´ı pr˚ ztr´atou ˇcichu. Na jejich l´eˇcbu se roˇcnˇe vynaloˇz´ı nemal´e n´alezem?“ u. L´eˇcba je nav´ıc ve vˇetˇsinˇe Anal´ yza bude narozd´ıl od vˇetˇsiny pˇredchoz´ıch studi´ı, mnoˇzstv´ı finanˇcn´ıch prostˇredk˚ pˇ r ´ ıpad˚ u pouze symptomatick´ a a doch´az´ı tak k ˇcast´ ym rekter´e ˇsly cestou kultivace bakteri´ı, prob´ıhat modern´ımi cidiv´ a m onemocnˇ e n´ ı a mnoho pˇ r ´ ıpad˚ u vyˇ z aduje opakometodami vyuˇz´ıvan´ ymi v molekul´ arn´ı biologii [2]. Bude se tedy jednat o anal´ yzu DNA, kter´ a bude z´ısk´ av´ana van´e chirurgick´e intervence [3]. z nosn´ıch stˇer˚ u a bioptick´ ych vzork˚ u z nosn´ıch sliznic. Bakteri´aln´ı DNA z´ıskan´ a ze vzork˚ u od pacient˚ u bude amplifikov´ana pomoc´ı specifick´ ych primer˚ u, urˇcen´ ych V pˇredchoz´ıch studi´ıch byl prok´az´an v´ yskyt r˚ uzn´ ych k amplifikaci 16S bakteri´ aln´ı DNA, jeˇz obsahuje jak druh˚ u mikrob˚ u v oblasti nosn´ ıch pr˚ u chod˚ u a vedlejˇ s ´ıch konzervativn´ı, tak variabiln´ı u ´seky. Ta bude nakonec dutin nosn´ ıch. Nebyla vˇ s ak prok´ a z´ a na pˇ r ´ ım´ a souvislost sekvenov´ana. U jednotliv´ ych pacient˚ u budou z´ aroveˇ n ym p˚ uvodcem ˇci skuprov´adˇena doplˇ nkov´a klinick´ a a paraklinick´ a vyˇsetˇren´ı. mezi vznikem onemocnˇen´ı a jednotliv´ pinou p˚ u vodc˚ u . Pˇ r ´ ıˇ c inou onemocnˇ e n´ ı je chronick´ y z´anˇet Tato data pak budou korelov´ ana k dat˚ um laboratorn´ım. sliznice dutin. Faktory, kter´ e zapˇ r ´ ıˇ c iˇ n uji tento z´anˇet, Dalˇs´ım v´ ystupem m˚ uˇze b´ yt i porovn´ an´ı z´ıskan´ ych dat jsou genetick´ e zmˇ e ny, alergie, infekce, pˇ r estavba sliznice s ostatn´ımi studiemi, a to jak s tˇemi realizovan´ ymi cestou kultivace, tak i s tˇemi proveden´ ymi metodami zaloˇzen´ ymi a poruchy imunomodulace [5]. U tohoto onemocnˇen´ı pak yˇsenou infiltraci sliznice eosinofily, neutrona anal´ yze nukleov´ ych kyselin. V tomto pˇr´ıpadˇe bude vyvol´avaj´ı zv´ jistˇe zaj´ımav´a pˇr´ıpadn´ a rozd´ılnost v´ ysledk˚ u v z´ avislosti fily, lymfocyty a makrof´agy, abnorm´aln´ı regulaci Th1, Th2 a regulaˇcn´ıch lymfocyt˚ u, poruchy slizniˇcn´ı, specina pouˇzit´e metodice. fick´e i nespecifick´e imunity, poˇskozen´ı epiteli´ı n´asledovan´e aberantn´ı pˇrestavbou, zmˇeny v pochodech ikosanoid˚ u Souˇ casn´ y stav pozn´ an´ı a fibr´ozu a otok. Ot´azkou je vˇsak pˇresn´ y mechanismus, kter´ y zapˇr´ıˇciˇ nuje vznik a pˇretrv´av´an´ı tˇechto zmˇen. uvodce inChronick´a rhinosinusitida je pomˇernˇe ˇcast´ ym one- A to zda-li pˇr´ıtomn´e bakterie slouˇz´ı jako p˚ mocnˇen´ım (5-15% populace v Evropˇe a USA) postihuj´ıc´ım fekce a z´anˇetu [4]. Dle souˇcasn´eho pozn´an´ı se tento vliv horn´ı cesty d´ ychac´ı. Pˇresn´ a etiologie je bohuˇzel nezn´am´a. pˇredpokl´ad´a. Avˇsak interpretace a porovn´an´ı pˇredeˇsl´ ych ysledky jsou ˇcasto z´avisl´e na Pravdˇepodobnˇe se vˇsak jedn´ a o onemocnˇen´ı multifak- studi´ı jsou znaˇcnˇe sloˇzit´e. V´ tori´aln´ı [1]. Je charakterizovan´ a jako lok´ aln´ı z´ anˇet posti- m´ıstˇe a zp˚ usobu odbˇeru materi´alu. Avˇsak i pˇres tyto huj´ıc´ı nosn´ı pr˚ uchody a paranaz´ aln´ı dutiny, jeˇz perzistuje rozd´ıly byla prok´az´ana rozd´ıln´a mikrobiologie u chroych a akutn´ıch rinosinusitid [3]. d´ele neˇz 12 t´ ydn˚ u. Chronick´e rhinosinusitidy rozdˇelujeme nick´ S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
ˇ Steffl M., Plz´ak J. – V´yskyt mikrobi´aln´ı fl´ ory nosn´ıch pr˚ uchod˚ u a jej´ı vliv na rozvoj chronick´ych rhinosinusitid
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı
ICD10: J33
Sekvenov´ an´ı nukleov´ ych kyselin
V´ ysledkem by mˇelo b´ yt zjiˇstˇen´ı, zda-li a jak se liˇs´ı os´ıdlen´ı sinonas´aln´ıho traktu u pacient˚ u s chronickou rhinosinusitidou a u zdrav´ ych jedinc˚ u a jak´ y mohou m´ıt tyto mikroorganizmy vliv na rozvoj tohoto onemocnˇen´ı, coˇz by ve v´ ysledku mohlo ovlivnit strategii l´eˇcby chronick´ ych rhinosinusitid. Dalˇs´ım zjiˇstˇen´ım by mˇelo b´ yt, zda-li a jak lze aplikovat nov´e molekul´ arn´ı metody do diagnostiky infekˇcn´ıch rhinosinusitid, jejich v´ yhody a nev´ yhody oproti standardn´ım kultivaˇcn´ım postup˚ um.
Definice: Urˇcov´an´ı pˇresn´eho poˇrad´ı nukleotid˚ u v molekule DNA. Synonyma: Molekul´arnˇe genetick´a anal´ yza Zdroj: https://en.wikipedia.org/wiki/ DNA_sequencing SNOMED CT: 117040002 MeSH: D008969
Podˇ ekov´ an´ı
ICD10: nenalezeno V´ yzkum podpoˇren SVV-2015-260158 projekt Karlovy Univerzity v Praze. Mikrobiom
Kl´ıˇ cov´ a slova
Definice: Sb´ırka genom˚ u mikrob˚ u v syst´emu. Synonyma: Mikrobiota
Chronick´ a rhinosinusitida Definice: Z´anˇet sinonaz´ aln´ıho traktu pˇretrv´ avaj´ıc´ı d´ele neˇz 12 t´ ydn˚ u.
Zdroj: microbe.net/2015/04/08/what-does -the-term-microbiome-mean-and-where-did -it-come-from-a-bit-of-a-surprise/
Synonyma: Rinitida, sinusitida
SNOMED CT: nenalezeno
Zdroj: www.hopkinsmedicine.org/ otolaryngology/specialty_areas/sinus_ center/conditions/sinusitis.html
MeSH: D064307 ICD10: nenalezeno
SNOMED CT: 40055000
Reference
MeSH: D012220, D012852
[1] Frank, D., Feazel, L., Bessesen, M., Price, C., Janoff, E., & Pace, N. (2010). The Human Nasal Microbiota and Staphylococcus aureus Carriage. PLoS ONE.
ICD10: J31.0, J32
Nosn´ı polypy Definice: Mˇekk´e, nebolestiv´e, nezhoubn´e nosn´ıch ˇci vedlejˇs´ıch dutin nosn´ıch.
v´ yr˚ ustky
[2] Stressmann, F., Rogers, G., Chan, S., Howarth, P., Harries, P., Bruce, K., & Salib, R. (2011). Characterization of bacterial community diversity in chronic rhinosinusitis infections using novel culture-independent techniques. Am J Rhinol Allergy American Journal of Rhinology and Allergy, 133-140.
Synonyma: V´ ychlipky nosn´ı sliznice
[3] Kennedy, D. (2012). Rhinology diseases of the nose, sinuses, and skull base. New York: Thieme.
Zdroj: www.mayoclinic.org/ diseases-conditions/nasal-polyps/basics/ definition/con-20023206
[4] Kato, A. (n.d.). Immunopathology of chronic rhinosinusitis. Allergology International, 121-130.
SNOMED CT: 52756005 MeSH: D009298
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
[5] Li, C., Shi, L., Yan, Y., Gordon, B., Gordon, W., & Wang, D. (2012). Gene Expression Signatures: A New Approach to Understanding the Pathophysiology of Chronic Rhinosinusitis. Curr Allergy Asthma Rep Current Allergy and Asthma Reports, 209-217.
55
56
Vesel´a Fl´ orov´a Z. et al. – L´eˇcba glaukomu za 1 Kˇc dennˇe – sen nebo realita?
L´ eˇ cba glaukomu za 1 Kˇ c dennˇ e – sen nebo realita? Zuzana Vesel´ a Fl´ orov´ a1 , Petr V´ yborn´ y2 , Silvia Siˇ c´ akov´ a2 , Jiˇr´ı Obenberger3 1 2
ˇ a republika Oˇcn´ı klinika JL Praha, s.r.o., Cesk´
´ ˇ a republika Oˇcn´ı klinika 1.LF UK a UVN – Vojensk´ a fakultn´ı nemocnice Praha, Cesk´ 3
ˇ a republika Radiodiagnostick´ a klinika FNB a 1. LF UK Praha, Cesk´
Kontakt: Zuzana Vesel´ a Fl´ orov´ a Oˇ cn´ı klinika JL Praha, s.r.o. ˇ a republika Adresa: V H˚ urk´ ach 1296/10, 158 00 Praha 5, Cesk´ E–mail:
[email protected]
C´ıle v´ yzkumu
pro analoga prostaglandin˚ u 5 Kˇc (5,0807 Kˇc) a skupina l´eˇciv´ ych pˇr´ıpravk˚ u v z´asadˇe terapeuticky zamˇeniteln´ ych ymi pˇr´ıpravky s obsahem fixn´ı kombinace timololu V posledn´ı dobˇe jsme se v oftalmologick´e praxi setkali s l´eˇciv´ ych analog˚ u m´a stanovenou z´akladn´ı s probl´emem v´ yrazn´eho omezen´ı preskripce fluorovan´ ych a prostaglandinov´ ´hradu za ODTD 5,5537 Kˇc [1]. chinolon˚ u E, OPH“ [3]. Omezen´ı vypl´ yvaj´ıc´ı ze spr´ avn´ıho u ” ´ rozhodnut´ı St´atn´ıho u ´ˇradu pro kontrolu l´eˇciv (SUKL) zaskoˇcila odbornou veˇrejnost. Situace, kter´ a nastala, n´as Aplikace v biomedic´ınˇ e motivovala k tomu, abychom se zaˇcali zaj´ımat o procesy ı souvisej´ıc´ı se stanoven´ım u ´hrady l´eˇciv´ ych pˇr´ıpravk˚ u. Po- a zdravotnictv´ kus´ıme se pˇribl´ıˇzit tuto problematiku a v n´ asleduj´ıc´ım ´ SUKL m´a za to, ˇze v´ ysledky hodnocen´ı n´akladov´e efektextu je pod´av´an v´ ybˇer z agendy t´ ykaj´ıc´ı se skupiny antitivity proveden´e v jin´ ych st´atech nelze pauˇs´alnˇe pˇrev´est glaukomatik. [4] ˇ jelikoˇz pˇri adaptaci zaKolik konkr´etnˇe odˇcerp´ a ze syst´emu zdravotn´ıho na zdravotnick´e prostˇred´ı CR, pojiˇstˇen´ı konzervativn´ı l´eˇcba pacienta s glaukomem? Jak hraniˇcn´ıch hodnocen´ı je nezbytn´e zohlednit charakter u, v´ yˇsi n´aklad˚ u, dejsou hrazeny oˇcn´ı kapky? Kolik vynaloˇz´ı st´ at na l´eˇcbu to- l´eˇcebn´e praxe, intenzitu ˇcerp´an´ı zdroj˚ hoto z´avaˇzn´eho onemocnˇen´ı jednotliv´ ymi terapeutick´ ymi finici c´ılov´e populace a jin´e kl´ıˇcov´e pˇredpoklady specifick´e ˇ [1]. pro prostˇred´ı bˇeˇzn´e klinick´e praxe v CR skupinami? Spr´avn´ı ˇr´ızen´ı prob´ıhaj´ı zpravidla ˇradu mˇes´ıc˚ u [2]. V´ ysledkem je pl´anovan´a roˇcn´ı u ´spora v ˇr´adu mili´on˚ u. Souˇ casn´ y stav pozn´ an´ı Na z´akladˇe v´ yˇse u ´hrady pˇredmˇetn´e skupiny v z´asadˇe ´ terapeuticky zamˇeniteln´ ych pˇr´ıpravk˚ u odhaduje SUKL, K dispozici odborn´e veˇrejnosti i laik˚ um jsou volnˇe ˇze dopad na prostˇredky veˇrejn´eho zdravotn´ıho pojiˇstˇen´ı pˇr´ıstupn´e webov´e str´anky st´ atn´ıch instituc´ı – v tomto bude n´asleduj´ıc´ı u ´spora n´aklad˚ u: pˇri l´eˇcbˇe betablok´atory ´ pˇr´ıpadˇe Ministerstva zdravotnictv´ı a SUKL [1, 2]. pˇribliˇznˇe 16 621 570 Kˇc, pˇri l´eˇcbˇe prostaglandiny pˇribliˇznˇe Tˇechto materi´al˚ u lze vyuˇz´ıt jako relevantn´ıho zdroje 60 938 769 Kˇc, pˇri l´eˇcbˇe kombinacemi prostaglandinu a tiinformac´ı z oblasti l´ekov´e politiky. Zde najdeme ce- mololu pˇribliˇznˇe 41 623 642 Kˇc. Odhad byl zpracov´an lou ˇradu podrobnost´ı z pr˚ ubˇehu jedn´ an´ı o stanoven´ı na z´akladˇe spotˇreb pˇr´ıpravk˚ u za rok 2012 a porovn´an´ım u ´hrad pˇr´ısluˇsn´ ych terapeutick´ ych skupin. Je zde moˇzno s u ´hradou platnou k 5. 1. 2013 [1, 2]. vyˇc´ıst nˇekolikaletou chronologii procesu stanoven´ı u ´hrady, Celkov´a pˇredpokl´adan´a u ´spora dosaˇzen´a s pomoc´ı n´amitky a pˇripom´ınky farmaceutick´ ych firem, kter´e se tˇechto tˇr´ı spr´avn´ıch ˇr´ızen´ı tedy dosahuje necel´ ych 120 snaˇz´ı upˇrednostnit sv˚ uj pˇr´ıpravek a z´ıskat pro nˇej pro- mili´on˚ u Kˇc. Jak´e je pl´anovan´e vyuˇzit´ı t´eto ˇc´astky? Jak´e centu´aln´ı bonifikaci, protoˇze nˇekter´e publikace mluv´ı dalˇs´ı ot´azky se nask´ ytaj´ı? Je moˇzn´e, aby u ´hrada ze zdrav jeho prospˇech. Jin´e publikace vˇsak naopak hovoˇr´ı ve votn´ıho pojiˇstˇen´ı u denn´ı l´eˇcby glaukomu byla u tak prospˇech pˇr´ıpravku jin´e firmy, a jsou tedy z logiky vˇeci rozˇs´ıˇren´e skupiny, jako jsou lok´aln´ı betablok´atory [3], uv´adˇeny konkurenc´ı. Z materi´ al˚ u lze i vyˇc´ıst, jak´ y je ocenˇena m´enˇe neˇz na hodnotu nejniˇzˇs´ı mince, kter´a je ´ ˇ v obˇehu? mechanismus hodnocen´ı tˇechto n´ amitek SUKL, a co je v CR z´akladem pro koneˇcn´e rozhodnut´ı o v´ yˇsi u ´hrady [1, 2]. Co bude n´asledovat? Jak´ y bude dopad dalˇs´ıch reviz´ı? Konkr´etn´ı v´ ysledek je alarmuj´ıc´ı – obvykl´ a denn´ı te- Budeme m´ıt nejniˇzˇs´ı n´aklady na jednotliv´e l´eky ze zdrarapeutick´a d´avka (ODTD) pro betablok´ atory m´ a stano- votn´ıho pojiˇstˇen´ı v Evropˇe? Jak´e bude portfolio oˇcn´ıch kaubec na trh venou z´akladn´ı u ´hradu necelou 1 Kˇc (0,9412 Kˇc), ODTD pek k l´eˇcbˇe glaukomu za nˇekolik let? Pˇrijdou v˚ S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Vesel´a Fl´orov´a Z. et al. – L´eˇcba glaukomu za 1 Kˇc dennˇe – sen nebo realita?
nov´e l´eky pˇri takto n´ızk´e u ´hradˇe a syst´emu jejich tvorby? Vyplat´ı se firmˇe n´ aklady spojen´e s registrac´ı l´eku? Kolik je pacient ochoten dopl´ acet v l´ek´ arnˇe? Jsou nˇekter´e oˇcn´ı kapky s neakceptovateln´ ymi doplatky prakticky obchodo” vateln´e“? Bude za st´ avaj´ıc´ıch podm´ınek ze strany drˇzitele registrace l´eku, kter´ y je nyn´ı souˇc´ ast´ı bˇeˇzn´e klinick´e praxe, z´ajem o jej´ı prodlouˇzen´ı? Kolik l´ek˚ u pˇrestane b´ yt k dispozici? Budou u n´ as distribuovat antiglaukomatika pouze generick´e firmy? Kolik jich bude? Snaha o u ´sporu n´ aklad˚ u na l´eˇciv´e pˇr´ıpravky hrazen´e ze zdravotn´ıho pojiˇstˇen´ı by mohla m´ıt pˇri pokraˇcuj´ıc´ım restriktivn´ım trendu i sv´e negativn´ı str´ anky. [1] V nepˇr´ıliˇs vzd´alen´em ˇcasov´em horizontu by mohlo doj´ıt k z´ uˇzen´ı terapeutick´eho portfolia, kter´e bude pacient˚ um a l´ekaˇr˚ um k dispozici pˇri l´eˇcbˇe z´ avaˇzn´ ych oˇcn´ıch chorob se soci´alnˇe ekonomick´ ym dopadem. Je tˇreba si uvˇedomit, ˇze praktick´ y v´ ysledek tˇechto opatˇren´ı se plnˇe projev´ı s latenc´ı nˇekolika let. Je d˚ uleˇzit´e o t´eto problematice diskutovat jiˇz nyn´ı.
Podˇ ekov´ an´ı
st´avaj´ı l´eˇciv´ ymi pˇr´ıpravky. Ve vhodn´em obalu a ve vhodn´ y okamˇzik pod´an´ı pacientovi se pak l´eˇciv´ y pˇr´ıpravek st´av´a l´ekem, kter´ y m˚ uˇze pˇr´ıznivˇe ovlivnit zdravotn´ı stav ˇci b´ yt pouˇzit k diagnostice onemocnˇen´ı. Zdroj: https://cs.wikipedia.org/wiki/L%C3%A9k SNOMED CT: 410942007 MeSH: nenalezeno ICD10: nenalezeno
´ Uhrada ze zdravotn´ıho pojiˇstˇ en´ı Definice: Vl´ada definuje rozsah p´eˇce hrazen´e z veˇrejn´eho zdravotn´ıho pojiˇstˇen´ı na z´akladˇe medic´ınsk´ ych krit´eri´ı, stupnˇe zdravotn´ıho postiˇzen´ı a v rozsahu moˇznost´ı veˇrejn´eho zdravotn´ıho pojiˇstˇen´ı. Definuje takt´eˇz ˇcasovou a m´ıstn´ı dostupnost zdravotn´ı p´eˇce a uloˇz´ı pl´atci tuto dostupnost pro pojiˇstˇence zajistit.
Tato pr´ace byla podpoˇrena projektem SVV-2015- Zdroj: http://www.aktualne.cz/wiki/politika\ 260158 Univerzity Karlovy v Praze. /zdravotnictvi/r$\sim$i:wiki:740/
Kl´ıˇ cov´ a slova
SNOMED CT: nenalezeno MeSH: nenalezeno
Glaukom ICD10: nenalezeno Definice: Glaukom je skupina klinicky odliˇsn´ ych one´ mocnˇen´ı r˚ uzn´e etiologie, kter´ a zp˚ usobuj´ı neuropatii SUKL zrakov´eho nervu a vedou k ireverzibiln´ımu poˇskozen´ı ˇ e zrakov´ ych funkc´ı. Je to onemocnˇen´ı multifaktori´aln´ı, Definice: St´atn´ı u ´stav pro kontrolu l´eˇciv je u ´ˇrad Cesk´ pro jehoˇz rozvoj je d˚ uleˇzit´ ym rizikov´ ym faktorem republiky, organizaˇcn´ı sloˇzka st´atu zˇr´ızen´a Miˇ e republiky, jej´ımˇz zv´ yˇsen´ y nitrooˇcn´ı tlak. nisterstvem zdravotnictv´ı Cesk´ ˇ e republice u ´kolem je dohl´ıˇzet na to, aby se v Cesk´ Zdroj: http://www.wikiskripta.eu/index.php/ pouˇz´ıvaly pouze jakostn´ı, bezpeˇcn´e a u ´ˇcinn´e l´eky, Glaukom jakoˇz i funkˇcn´ı a bezpeˇcn´e zdravotnick´e pom˚ ucky. SNOMED CT: 23986001
Zdroj: http://cs.wikipedia.org/wiki/SUKL
MeSH: C11.525.381
SNOMED CT: nenalezeno
ICD10: H40-H42
MeSH: nenalezeno
L´ ek
ICD10: nenalezeno
Definice: L´ek je l´eˇcivo upraven´e do definitivn´ı podoby, v jak´e se pouˇz´ıv´ a a pod´ av´ a pacientovi (ˇclovˇeku nebo zv´ıˇreti). Pojem l´ek“ je definov´ an ve far” makologii (zab´ yv´ a se l´eky). Na rozd´ıl od pojm˚ u l´eˇcivo“, l´eˇciv´ a l´ atka“ a l´eˇciv´ y pˇr´ıpravek“ nem´a ” ” ” l´ek“ definici v z´ akonˇe. Vˇsechna l´eˇciva (l´eˇciv´e l´atky ” a l´eˇciv´e pˇr´ıpravky) jsou potenci´ aln´ımi l´eky, kter´ ymi se st´avaj´ı v okamˇziku, kdy jsou spr´ avn´ ym zp˚ usobem pod´any pacientovi. Proces je vyj´ adˇren tzv. z´akonem o vzniku l´eku, kter´ y ˇr´ık´ a, ˇze na poˇc´ atku jsou l´eˇciv´e l´atky, kter´e se v pr˚ ubˇehu technologick´ ych proces˚ u a po sm´ıch´ an´ı s vhodn´ ymi pomocn´ ymi l´atkami
Antiglaukomatikum
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Definice: Antiglaukomatikum je l´eˇcivo uˇz´ıvan´e k l´eˇcbˇe zelen´eho z´akalu. Zdroj: http://www.olecich.cz/slovnik /antiglaukomatikum SNOMED CT: 419886007 MeSH: nenalezeno ICD10: nenalezeno
57
58
Vesel´a Fl´ orov´a Z. et al. – L´eˇcba glaukomu za 1 Kˇc dennˇe – sen nebo realita?
Reference
hodnut´ı
[1] http://www.mzcr.cz, data dostupn´ az http://www.mzcr.cz/leky z 29.12.2014
[3] V´ yborn´ y P., Siˇ ca ´kov´ a S., Dohnalov´ a P., Feˇrtek M., Doleˇ zal ˇ T.: Terapie glaukomu – aktu´ aln´ı pˇrehled dat a informac´ı. Ces. a Slov. Oftal. 69, 2013, 3: 118-126.
[2] http.//www.sukl.cz, data dostupn´ a pod ˇ c´ısly Spr´ avn´ıch roz-
[4] http://www.prolekare.cz/glakom-novinky/
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Vlas´akov´a M. et al. – Telemonitoring z´akladn´ıch terapeutick´ych prvk˚ u l´eˇcby diabetes mellitus . . .
Telemonitoring z´ akladn´ıch terapeutick´ ych prvk˚ u l´ eˇ cby diabetes mellitus a moˇ znosti jejich hodnocen´ı Martina Vlas´ akov´ a1 , Miroslav Muˇ zn´ y1,2 , Jan Muˇ z´ık3 1 2 3
ˇ a republika 1. l´ekaˇrsk´ a fakulta Univerzity Karlovy, Praha, Cesk´
ˇ eho vysok´eho uˇcen´ı technick´eho, Kladno, Cesk´ ˇ a republika Fakulta biomedic´ınsk´eho inˇzen´yrstv´ı Cesk´
ˇ a republika Centrum podpory aplikaˇcn´ıch v´ystup˚ u a spin-of firem Dˇekan´ atu 1. l´ekaˇrsk´e fakulty Univerzity Karlovy, Praha, Cesk´
Kontakt: Martina Vlas´ akov´ a 1. l´ ekaˇrsk´ a fakulta Univerzity Karlovy v Praze Adresa: Kateˇrinsk´ a 32, 121 08 Praha 2 E–mail:
[email protected]
C´ıle v´ yzkumu Hlavn´ım c´ılem v´ yzkumu je zefektivnˇen´ı p´eˇce o diabetick´e pacienty s vyuˇzit´ım modern´ıch technologi´ı, jejichˇz pouˇzit´ı umoˇzn´ı zlepˇsen´ı metabolick´e kontroly nemoci a zv´ yˇsen´ı kvality ˇzivota pacient˚ u. Tohoto c´ıle bude dosaˇzeno prostˇrednictv´ım v´ yvoje telemedic´ınsk´eho syst´emu (Obr´azek 1), kter´ y integruje tˇri hlavn´ı technologie: modern´ı aplikaci pro chytr´e mobiln´ı telefony zaznamen´avaj´ıc´ı ˇradu parametr˚ u t´ ykaj´ıc´ıch se pˇredevˇs´ım pˇr´ıjmu potravy a d´avek inzul´ınu ˇci jin´ ych antidiabetik, aktivity tracker pro sn´ım´an´ı a hodnocen´ı fyzick´e aktivity pacienta a glukometr umoˇzn ˇuj´ıc´ı online pˇrenos namˇeˇren´ ych hodnot.
Souˇc´ast´ı syst´emu je internetov´ y server zaznamen´avaj´ıc´ı a vyhodnocuj´ıc´ı tato pacientsk´ a data, jeˇz do nˇej budou automaticky pˇrenesena chytr´ ym telefonem v re´aln´em ˇcase. Pro kontrolu glyk´emie umoˇzn ˇuje telemedic´ınsk´ y syst´em nav´ıc synchronizaci dat z kontinu´aln´ıho monitorovac´ıho syst´emu. Dalˇs´ım moˇzn´ ym prvkem syst´emu jsou i inteligentn´ı hodinky, do kter´ ych lze nahr´at aplikaci, kter´ a komunikuje s aplikac´ı v mobiln´ım telefonu a umoˇzn ˇuje uˇzivateli obousmˇern´e sd´ılen´ı dat.
je z´aroveˇ n dostateˇcnˇe motivovat pacienty dan´e veliˇciny mˇeˇrit (bolestivost, ˇcasov´a nebo technick´a n´aroˇcnost, n´ızk´a edukace, nedostateˇcn´a motivace, n´ızk´ y komfort). Pˇritom kvalitn´ı data a hlavnˇe jejich vˇcasn´e vyhodnocen´ı jsou z´akladem u ´spˇeˇsn´e kompenzace nemoci. Pˇri kompenzaci diabetu je st´ale v´ıce vyuˇz´ıvanˇejˇs´ım n´astrojem telemonitoring. S pomoc´ı telemonitoringu z´ısk´avaj´ı l´ekaˇri pˇresn´a a spolehliv´a data v re´aln´em ˇcase [2]. Vzd´alen´a monitorace ovlivˇ nuje postoje a chov´an´ı pacient˚ u a potenci´alnˇe tak zlepˇsuje jejich zdravotn´ı stav [3]. Telemonitoring m´a pro pacienty motivaˇcn´ı a vzdˇel´avac´ı efekt a l´ekaˇri umoˇzn ˇuje rychlejˇs´ı vyhodnocen´ı v´ ysledk˚ u l´eˇcby. Telemedic´ınsk´ y syst´em pˇrin´aˇs´ı zv´ yˇsen´ı kvality zdravotn´ı p´eˇce a z´aroveˇ n sniˇzuje potˇrebu vyuˇz´ıvat zdravotn´ıch sluˇzeb, pozitivnˇe tak ovlivˇ nuje v´ yˇsi vynakl´adan´ ych finanˇcn´ıch prostˇredk˚ u na l´eˇcbu. Stabilizovan´ y stav pacienta m´a kladn´ y vliv na jeho celkovou spokojenost a ˇzivotn´ı u ´roveˇ n. Telemonitoring je zaloˇzen na komunikaci odes´ılatele s pˇr´ıjemcem v re´aln´em ˇcase, kdy je umoˇznˇena okamˇzit´a reakce zdravotn´ıka na podnˇet pacienta [4]. Pacient vyuˇz´ıv´a pro kontrolu biologick´ ych parametr˚ u interaktivn´ı zaˇr´ızen´ı, kter´e pomoc´ı internetu pˇren´aˇs´ı data k l´ekaˇri. Ten je vyhodnocuje a vol´ı dalˇs´ı postup.
Souˇ casn´ y stav pozn´ an´ı
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı
Diabetes mellitus souhrnnˇe oznaˇcuje chronick´e onemocnˇen´ı metabolismu, kter´e postihuje zejm´ena sacharidy. Pˇri kompenzaci onemocnˇen´ı se uplatˇ nuje ˇrada faktor˚ u. Tˇemi nejv´ yznamnˇejˇs´ımi jsou mnoˇzstv´ı pˇrij´ıman´ ych sacharid˚ u, d´avka inzul´ınu a m´ıra fyzick´e aktivity [1]. Tyto parametry lze kontrolovat pomoc´ı dostupn´ ych technick´ ych prostˇredk˚ u. Probl´emem vˇsak b´ yv´ a vyhodnocen´ı tˇechto veliˇcin pro jejich velk´e mnoˇzstv´ı a nesourodost. Obt´ıˇzn´e
Navrˇzen´e technick´e ˇreˇsen´ı pro online sbˇer dat a centralizaci sledovan´ ych parametr˚ u pro jednotliv´e diabetick´e pacienty poskytne jim i jejich l´ekaˇr˚ um pˇrehlednˇejˇs´ı a utˇr´ıdˇen´ y z´aznam sledovan´ ych veliˇcin. Syst´em umoˇzn´ı zaznamen´avat t´emˇeˇr v re´aln´em ˇcase trendy ve v´ yvoji hladiny glyk´emie a z´aroveˇ n pozorovat m´ıru z´avislosti v´ yvoje hladiny glyk´emie na m´ıˇre fyzick´e aktivity. Souˇc´ast´ı telemedic´ınsk´eho syst´emu je pˇredzpracov´an´ı, utˇr´ıdˇen´ı dat
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
59
60
Vlas´akov´a M. et al. – Telemonitoring z´akladn´ıch terapeutick´ych prvk˚ u l´eˇcby diabetes mellitus . . .
Obr´ azek 1: N´ avrh telemedic´ınsk´eho syst´emu.
a moˇznost proveden´ı statistick´eho porovn´ an´ı namˇeˇren´ ych hodnot.
Podˇ ekov´ an´ı
Tato pr´ace byla podpoˇrena projektem SVV-2015Pˇr´ınosem pro pacienta je poskytnut´ı pˇr´ım´e zpˇetn´e vazby odr´aˇzej´ıc´ı vliv jeho aktu´ aln´ıho ˇzivotn´ıho stylu na 260158 Univerzity Karlovy v Praze. pr˚ ubˇeh nemoci. Pacient z´ aroveˇ n z´ısk´ a lepˇs´ı kontrolu nad sv´ ym onemocnˇen´ım, kter´ a potenci´ alnˇe zvyˇsuje motivaci Kl´ ıˇ cov´ a slova dodrˇzovat stanovenou terapii. L´ekaˇri syst´em umoˇzn´ı bezprostˇredn´ı kontrolu kompenzace onemocnˇen´ı u pacienta Diabetes Mellitus s moˇznost´ı vˇcasn´e u ´pravy nastaven´e l´eˇcby. Navrˇzen´e ˇreˇsen´ı pˇrispˇeje ke sn´ıˇzen´ı vzniku negativn´ıch Definice: Skupina metabolick´ ych chorob, kter´e se provliv˚ u ˇspatnˇe kompenzovan´eho onemocnˇen´ı (vznik hypojevuj´ı u pacienta vysokou hladinou krevn´ıho cukru. glyk´emie v d˚ usledku n´ızk´e hladiny glyk´emie a komorbiVznik´a v d˚ usledku nedostatku inzulinu, jeho nedodit v d˚ usledku ˇspatn´e kompenzace vysok´e hladiny gluk´ozy stateˇcn´eho u ´ˇcinku, nebo kombinac´ı oboj´ıho. v krvi). ´ Synonyma: Cukrovka, Uplavice cukrov´a Vyuˇzit´ı telemedic´ınsk´eho syst´emu tak potenci´alnˇe m˚ uˇze v´est ke sn´ıˇzen´ı n´aklad˚ u vynakl´ adan´ ych na l´eˇcbu pa- Zdroj: Velk´ y l´ekaˇrsk´ y slovn´ık. http://lekarske. cienta a zv´ yˇsen´ı jeho kvality ˇzivota. slovniky.cz (pˇr´ıstup 28. 6. 2014) S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Vlas´akov´a M. et al. – Telemonitoring z´akladn´ıch terapeutick´ych prvk˚ u l´eˇcby diabetes mellitus . . .
SNOMED CT: 191044006 MeSH: D003920
Synonyma: Ambulatory continuous glucose monitoring of interstitial tissue fluid
ICD10: E10-E14
Zdroj: Edelsberger T. Encyklopedie pro diabetiky, 1st ed. Praha: Maxdorf; 2009.
Telemedic´ına
SNOMED CT: 439926003
Definice: Poskytov´ an´ı zdravotnick´ ych sluˇzeb prostˇred- MeSH: nenalezeno nictv´ım vzd´ alen´ ych telekomunikac´ı (zahrnuje interICD10: nenalezeno aktivn´ı poradenstv´ı a diagnostick´e sluˇzby). Synonyma: eHealth, mHealth, Telehealth
Glyk´ emie
Zdroj: National Library of Medicine – Medical SubDefinice: Koncentrace hladiny gluk´ozy v krvi. Je ject Headings. http://www.nlm.nih.gov (pˇr´ıstup udrˇzov´ana v pomˇernˇe st´al´em rozmez´ı, protoˇze 26. 6. 2015) pˇr´ısun cukr˚ u je d˚ uleˇzit´ y pro ˇradu org´an˚ u, zejm. mozek. Pokles pod doln´ ı hranici normy se naz´ yv´a hySNOMED CT: 448337001 poglykemie a zv´ yˇsen´a hladina se oznaˇcuje jako hyMeSH: D017216 perglykemie. ICD10: nenalezeno
Synonyma: Blood Sugar
Mobiln´ı aplikace
Zdroj: Velk´ y l´ekaˇrsk´ y slovn´ık. http://lekarske. slovniky.cz (pˇr´ıstup 25. 6. 2015)
Definice: Poˇc´ıtaˇcov´e programy nebo software nainstalov´any do mobiln´ıch elektronick´ ych zaˇr´ızen´ı, kter´a SNOMED CT: 365812005 podporuj´ı ˇsirokou ˇsk´ alu funkc´ı a pouˇzit´ı, kter´e zahr- MeSH: D001786 nuj´ı televizi, telefon, video, hudbu, zpracov´an´ı textu, a internetov´ ych sluˇzeb. ICD10: R73 Synonyma: Mobile Apps,Portable Electronic Applications Zdroj: National Library of Medicine – Medical Subject Headings. http://www.nlm.nih.gov (pˇr´ıstup 26. 6. 2015) SNOMED CT: nenalezeno MeSH: D063731 ICD10: nenalezeno
Kontinu´ aln´ı mˇ eˇren´ı glyk´ emie Definice: Oznaˇcuje zp˚ usob mˇeˇren´ı krevn´ıho cukru (glyk´emie) v kr´ atk´ ych intervalech (5 minut) v pr˚ ubˇehu jednoho a v´ıce dn´ı pomoc´ı speci´aln´ıho ´ senzoru um´ıstˇen´eho v podkoˇz´ı. Udaje se pˇren´aˇsej´ı bud’ pˇr´ımo na displej, nebo po pˇrenosu dat do poˇc´ıtaˇce.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Reference [1] Jaana, M. Home telemonitoring of patients with diabetes: a systematic assessment of observed effects. Journal of Evaluation in Clinical Practice 2006; (13): 242–253. http://www.ncbi.nlm.nih.gov/pubmed/17378871 (accessed 10. December 2012). [2] Mignerat M, Lapointe L, Vedel I. Using telecare for diabetic patients: A mixed systematic review. Health Policy and Technology. 2014;3(2):90-112. [3] Glasgow R.E.. D net diabetes self management program: long-term implematation, outcomes, and generatization results. Preventive medicine 2003; (36): 410-419. http://www.ncbi.nlm.nih.gov/pubmed/12649049 (accessed 10. December 2012). [4] Stone R.A. The Diabetes Telemonitoring Study Extension: an exploratory randomized comparison of alternative interventions to maintain glycemic control after withdrawal of diabetes home telemonitoring. Journal of the American Medical Informatics Association 2012; (66): 973–979. http://www.ncbi.nlm.nih.gov/pubmed/22610495 (accessed 5. December 2012).
61
62
Vondruˇskov´a L. – V´yhody vyuˇzit´ı virtu´aln´ıho pacienta v nel´ekaˇrsk´ych zdravotnick´ych oborech
V´ yhody vyuˇ zit´ı virtu´ aln´ıho pacienta v nel´ ekaˇrsk´ ych zdravotnick´ ych oborech Lenka Vondruˇskov´ a1 1
ˇ a republika Neurochirurgick´ a klinika – JIP, Fakultn´ı nemocnice Plzeˇ n, Cesk´
Kontakt: Lenka Vondruˇskov´ a ˇ a Republika Neurochirurgick´ a klinika – JIP, Fakultn´ı nemocnice Plzeˇ n, Cesk´ Adresa: Dr. E. Beneˇse 13, 301 00 Plzeˇ n E–mail:
[email protected]
C´ıle v´ yzkumu Virtu´aln´ı pacient (VP) je pojem, kter´ y je nejednoznaˇcnˇe ch´ap´an. Nˇekteˇr´ı si pod n´ım pˇredstavuj´ı umˇele vytvoˇren´ y sofistikovan´ y model napodobuj´ıc´ı pacienta, jin´ı jej ch´apou jako poˇc´ıtaˇcov´ y software. Asociace americk´ ych l´ekaˇrsk´ ych fakult definuje VP jako specifick´ y typ poˇc´ıtaˇcov´eho programu, kter´ y simuluje re´aln´e klinick´e sc´en´ aˇre, vede studenty v roli poskytovatele zdravotn´ı p´eˇce pˇri z´ısk´ av´ an´ı anamn´ezy, n´ asledn´em klinick´em vyˇsetˇren´ı a pˇri stanoven´ı diagn´ ozy a l´eˇcebn´eho pl´anu. [1] Cook a Triola popisuj´ı VP jako klinick´e sc´en´ aˇre, kter´e jsou pˇrehr´av´any na poˇc´ıtaˇcov´e obrazovce. Studenti hodnot´ı pacienta typov´an´ım nebo v´ ybˇerem nab´ızej´ıc´ıch se odpovˇed´ı s moˇznost´ı doplnˇen´ı napˇr. v´ ysledk˚ u laboratorn´ıch test˚ u. Poˇc´ıtaˇc student˚ um nab´ız´ı odpovˇedi ˇci doplˇ nuj´ıc´ı informace o zdravotn´ım stavu pacienta. Od student˚ u se oˇcek´av´a nastaven´ı diagn´ ozy a l´eˇcebn´ y pl´ an. VP by mˇel b´ yt urˇcen a vyuˇz´ıv´an k podpoˇre klinick´ ych rozhodovac´ıch dovednost´ı. [2] Dle Hursta je u ´ˇcelem VP vzdˇel´ av´ an´ı student˚ u a zdravotnick´ ych profesion´al˚ u prostˇrednictv´ım poˇc´ıtaˇc˚ u, kter´e simuluj´ı re´alnou situaci ze zdravotnick´eho prostˇred´ı s vyuˇzit´ım virtu´aln´ıho instruktora a zprostˇredkov´an´ım zpˇetn´e vazby. [3] C´ılem v´ yzkumu je tedy vytvoˇren´ı poˇc´ıtaˇcov´ ych klinick´ ych sc´en´aˇr˚ u pro nel´ekaˇrsk´e zdravotnick´e pracovn´ıky a n´asledn´e testov´ an´ı efektivity tˇechto sc´en´ aˇr˚ u ve vzdˇel´av´an´ı.
Souˇ casn´ y stav pozn´ an´ı D´a se pˇredpokl´adat, ˇze pokud je virtu´ aln´ı klinick´ y sc´en´aˇr obrazem re´aln´e klinick´e situace a re´ aln´eho v´ ybˇeru diagnostick´ ych a terapeutick´ ych krok˚ u, m˚ uˇze b´ yt VP uˇziteˇcnou podp˚ urnou pom˚ uckou pro testov´ an´ı znalost´ı a pˇri rozhodov´an´ı. Stevens upozorˇ nuje, ˇze budoucnost vyuˇz´ıv´an´ı VP z´avis´ı na v´ yvoji a hodnotic´ıch metod´ach,
kter´e povedou ke zv´ yˇsen´ı vyuˇzit´ı virtu´aln´ıch klinick´ ych sc´en´aˇr˚ u. [4] V´ yhodou virtu´aln´ıch klinick´ ych sc´en´aˇr˚ u je mimo jin´e sn´ıˇzen´ı rizik na rozd´ıl od re´aln´eho pacienta. T´ımto t´ematem se zab´ yvali autoˇri jako Eagles, kter´ y upozorˇ noval na nev´ yhodnost re´aln´eho pacienta, [5] nebo Zary, kter´ y se zm´ınil, ˇze u VP je na rozd´ıl od re´aln´eho pacienta chybov´an´ı povoleno. [6] O v´ yhodnosti VP ve spojen´ı se sn´ıˇzen´ım rizik v˚ uˇci re´aln´ ym pacient˚ um pojedn´avaj´ı i Gordon a Stevens. [7, 4] Pro zajiˇstˇen´ı kvality vytvoˇren´eho VP lze vyuˇz´ıt standard˚ u International Organization for Standardization (ISO). [8, 9] Kvalitou se zab´ yv´a i spoleˇcnost The European Committee for Standardization (CEN). [10] Rozvojem a podporou technologick´ ych standard˚ u se zab´ yv´a neziskov´a mezin´arodn´ı skupina MedBuiquitous. Je tˇreba promyslet tvorbu nov´ ych forem VP a vyuˇzit´ı st´avaj´ıc´ıch podp˚ urn´ ych vzdˇel´avac´ıch program˚ u s pˇrihl´ednut´ım k u ´spoˇre finanˇcn´ıch n´aklad˚ u.
Uplatnˇ en´ı v biomedic´ınˇ e a zdravotnictv´ı Z´ajem o rozvoj podp˚ urn´ ych vzdˇel´avac´ıch opor pro zdravotnick´e profesion´aly a moˇznost vyuˇzit´ı virtu´aln´ıch klinick´ ych sc´en´aˇr˚ u v nel´ekaˇrsk´ ych zdravotnick´ ych oborech souvis´ı se strategick´ ymi dokumenty eHealth. (Akˇcn´ı pl´an eHealth, i2010 Evropsk´a informaˇcn´ı spoleˇcnost pro r˚ ust a zamˇestnanost, Akˇcn´ı pl´an eHealth (eHAP) 2012–2020). Z´ajem je patrn´ y i ze strany Ministerstva zdravotnictv´ı ˇ (MZ CR) ˇ v projektu Prohlubov´an´ı a zvyˇsov´an´ı u CR ´rovnˇe odborn´ ych znalost´ı (l´ekaˇr˚ u a nel´ekaˇr˚ u). Nel´ekaˇrˇst´ı zdravotniˇct´ı pracovn´ıci maj´ı ze z´akona ustanovenu povinnost se vzdˇel´avat dle ustanoven´ı § 67 z´akona ˇc. 96/2004 Sb., o podm´ınk´ach z´ısk´av´an´ı a uzn´av´an´ı zp˚ usobilosti k v´ ykonu nel´ekaˇrsk´ ych zdravotnick´ ych povol´an´ı a k v´ ykonu ˇcinnost´ı souvisej´ıc´ıch s poskytov´an´ım zdravotn´ı p´eˇce a o zmˇenˇe nˇekter´ ych souvisej´ıc´ıch z´akon˚ u S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Vondruˇskov´a L. – V´yhody vyuˇzit´ı virtu´aln´ıho pacienta v nel´ekaˇrsk´ych zdravotnick´ych oborech
(z´akon o nel´ekaˇrsk´ ych zdravotnick´ ych povol´an´ıch) ve SNOMED CT: nenalezeno znˇen´ı pozdˇejˇs´ıch pˇredpis˚ u. S t´ım souvis´ı dvˇe ot´ azky. M˚ uˇze VP sn´ıˇzit fluktu- MeSH: D003198 aci zamˇestnanc˚ u ve zdravotnictv´ı? M˚ uˇze pomoci k rychICD10: nenalezeno lejˇs´ımu zaˇskolen´ı nov´ ych zamˇestnanc˚ u ve zdravotnictv´ı?
Podˇ ekov´ an´ı Tato pr´ace byla podpoˇrena projektem SVV-2015260158 Univerzity Karlovy v Praze.
Kl´ıˇ cov´ a slova Virtu´ aln´ı pacient
Klinick´ y Definice: T´ ykaj´ıc´ı se kliniky. Protikladem m˚ uˇze b´ yt teoretick´ y (chirurgie je k. oborem medic´ıny, anatomie oborem teoretick´ ym) nebo ambulantn´ı (k. a ambulantn´ı ˇca´st nemocnice). Zdroj: Vokurka, M., Hugo, J. a kol. Praktick´ y slovn´ık medic´ıny. Praha: Maxdorf, 2011, str. 234
SNOMED CT: 58147004 Definice: Interaktivn´ı poˇc´ıtaˇcov´ a simulace re´aln´eho klinick´eho sc´en´ aˇre pro u ´ˇcely medic´ınsk´eho tr´eninku, MeSH: D015510 edukace nebo vyˇsetˇren´ı. Zdroj: Ellaway, R., Candler, C., Greene, P., Smothers, ICD10: Z006 V., 2006. An Architectural Model forMedBiquitous Riziko VPs. MedBiquitous, Baltimore, MD SNOMED CT: nenalezeno MeSH: nenalezeno ICD10: nenalezeno
Vzdˇ el´ av´ an´ı Definice: Komplexn´ı pojem vyjadˇruj´ıc´ı, za jak´ ych podm´ınek vedou urˇcit´e vstupy vzdˇel´ avac´ıch proces˚ u k urˇcit´ ym v´ ystup˚ um. Vstupy pˇredstavuj´ı faktory dan´e charakteristikami subjekt˚ u a obsahu vzdˇel´av´an´ı, podm´ınky jsou vytv´ aˇreny charakteristikami proces˚ u v´ yuky a v´ ystupy jsou vzdˇel´avac´ı v´ ysledky a efekty vzdˇel´ av´ an´ı. Mˇeˇren´ı a vyhodnocov´an´ı efektivnosti vzdˇel´ av´ an´ı v praxi je sloˇzit´e, i kdyˇz v´ yzkum v t´eto oblasti je intenzivnˇe rozv´ıjen. Zdroj: Kol´aˇr, Z. a kol. V´ ykladov´ y slovn´ık z pedagogiky. Praha: Grada, 2012. ISBN: 978-80-247-3710-2. p. 179 SNOMED CT: 2760311006 MeSH: Q000193 ICD10: Z71.9
Simulace Definice: Napodoben´ı jednotliv´ ych dˇej˚ u nebo chov´an´ı cel´eho syst´emu. Synonyma: Simulov´ an´ı Zdroj: Nˇemeˇcek, M. a kol. Struˇcn´ y slovn´ık didaktick´e techniky a uˇcebn´ıch pom˚ ucek. Praha: St´atn´ı pedagogick´e nakladatelstv´ı, 1985, p. 82 S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Definice: Pravdˇepodobnost vzniku nepˇr´ızniv´e ud´alosti, napˇr. vzniku onemocnˇen´ı. Riziko se zjiˇst’uje statistick´ ymi, resp. epidemiologick´ ymi metodami, napˇr´ıklad dlouhodob´ ym sledov´an´ım v´ yskytu urˇcit´eho onemocnˇen´ı ve vybran´e skupinˇe osob, na urˇcit´em u ´zem´ı, v z´avislosti na r˚ uzn´ ych okolnostech apod. Zdroj: Vokurka, M., Hugo, J. a kol. Praktick´ y slovn´ık medic´ıny. Praha: Maxdorf, 2011, str. 403 SNOMED CT: 129839007 MeSH: D012306 ICD10: nenalezeno
Reference [1] Association of American Medical Colleges, 2007. Effective Use of Educational Technology in Medical Education: Summary Report of the 2006 AMC Colloquium on Educational Technology. AMC, Washington DC [2] Cook, D.A., Triola, M.M., 2009. VPs: a critical literature review and proposed next steps. Medical Education 43, 303e311 [3] Hurst, M.H., Marks-Maran, D. Using a virtual patient aktivity to teach nurse prescribing. Nurse Education in Practice 2011; 11:192-198 [4] Stevens, A., Hernandez, J., Johnsen, K., Dickerson, R., Raij, A., Harrison, C., DiPietro, M., Allen, B., Ferdig, R., Foti, S., Jackson, J., Shin, M., Cendan, J., Watson, R., Duerson, M., Lok, B., Cohen, M., Wagner, P., Lind, D.S., 2006. The use of VPs to teach medical students history taking and communication skills. American Journal of Surgery 191, 806e811 [5] Eagles, J., Calder, S., Nicoll, K., Sclare, P.D., 2001. Using simulated patients in education about alcohol misuse. Academic Medicine 76 (4), 395
63
64
Vondruˇskov´a L. – V´yhody vyuˇzit´ı virtu´aln´ıho pacienta v nel´ekaˇrsk´ych zdravotnick´ych oborech
[6] Zary, N., Johnson, G., Boberg, J., Fors, U., 2006. Development, implementation and pilot evaluation of a web-based VP case simulation environment e web-SP. BioMed Central Medical Education 6,10. At. http://www.pubmedcentral.nih.gov/ articlerender.fcgi?tool-pubmed&pubmedid-16504041 (accessed 10.07.09) [7] Gordon, J.A., Wilkerson, W.M., Schafer, D.W., Armstrong, E.G., 2001. Practicing medicine without risk: students’ and educators’ responses to high-fidelity patient simulation. Academic Medicine 76 (5), 469e472
[8] ISO/IEC 19796-1. Information technology – Learning, education and training – Quality management, assurance and metrics: Part 1: General approach. Switzerland: ISO/IEC 2005 [9] ISO/IEC 19796-3. Information technology – Learning, education and training – Quality management, assurance and metrics: Part 3: Reference methods and metrics. Switzerland: ISO copyright office, 2009 [10] CEN/ISSS CWA 14644: Quality Assurance and Guidelines. Brussels: CEN/ISSS, 2003
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Zv´ara K. a kol. – Pˇredzpracov´an´ı l´ekaˇrsk´ych zpr´av pro extrakci informac´ı
Pˇredzpracov´ an´ı l´ ekaˇrsk´ ych zpr´ av pro extrakci informac´ı Karel Zv´ ara1 , Marie Tomeˇ ckov´ a2 , Vojtˇ ech Sv´ atek3 , Jana Zv´ arov´ a1 1
´ ˇ Ustav hygieny a epidemiologie, 1. l´ekaˇrsk´ a fakulta UK, Praha, CR 2
3
ˇ EuroMISE Mentor Association, Praha, CR
ˇ Praha, CR ˇ Katedra informaˇcn´ıho a znalostn´ıho inˇzen´yrstv´ı, Fakulta informatiky a statistiky, VSE
Kontakt: Karel Zv´ ara ´ Ustav hygieny a epidemiologie, 1. l´ ekaˇrsk´ a fakulta UK ˇ Adresa: Studniˇ ckova 7, 121 08 Praha 2, CR E–mail:
[email protected]
C´ıle v´ yzkumu Extrakce informac´ı z l´ekaˇrsk´ ych zpr´ av je d˚ uleˇzit´ ym krokem pro kvalitn´ı a efektivn´ı rozhodov´ an´ı ve zdravotnictv´ı. Hlavn´ım v´ yzkumn´ ym c´ılem je vyvinout metodu pro pˇr´ıpravu l´ekaˇrsk´ ych zpr´ av pro extrakci informac´ı z voln´eho textu do strukturovan´e formy. Strukturovan´ a informace m˚ uˇze b´ yt d´ale uloˇzena v syst´emech pro elektronick´ y zdravotn´ı z´aznam.
Souˇ casn´ y stav pozn´ an´ı Pro z´ısk´an´ı plnˇe pouˇziteln´e klinick´e informace z l´ekaˇrsk´ ych zpr´ av je zapotˇreb´ı takovou informaci uchov´avat v plnˇe strukturovan´e formˇe. Mnoho vˇedc˚ u se jiˇz zab´ yvalo probl´emem z´ısk´ an´ı strukturovan´e formy z voln´ ych text˚ u. Je zˇrejm´e, ˇze proces z´ısk´ av´ an´ı strukturovan´e informace z voln´eho textu nez´ avis´ı jen na jazyku [1], ale tak´e na m´ıstn´ıch poˇzadavc´ıch (vˇc. legislativn´ıch) a na tom, jak´a osoba dokumentaci poˇrizuje. Nejednoznaˇcnost v´ yraz˚ u zd˚ uraznil ˇcl´ anek [2], kdyˇz citov´ an´ım dalˇs´ıch zdroj˚ u uvedl, ˇze polovinˇe bˇeˇznˇe uˇz´ıvan´ ych zkratek v oboru uˇsn´ı, nosn´ı a krˇcn´ı chirurgie nerozum´ı v´ıce neˇz 90 % zaˇc´ınaj´ıc´ıch l´ekaˇr˚ u z jin´ ych obor˚ u. V minulosti se podaˇrilo dos´ahnout urˇcit´ ych u ´spˇech˚ u pˇri extrahov´ an´ı numerick´ ych u ´daj˚ u [3] (napˇr. hodnot krevn´ıho tlaku).
Aplikace v biomedic´ınˇ e a zdravotnictv´ı
vyuˇzit´ım m˚ uˇze b´ yt mˇeˇren´ı efektivity a kvality zdravotn´ı p´eˇce ˇci vyuˇzit´ı pro on-line n´apovˇedu a upozorˇ nov´an´ı pˇri poskytov´an´ı p´eˇce. Byl vyvinut syst´em pro zpracov´an´ı l´ekaˇrsk´ ych zpr´av, kter´ y je pˇr´ıstupn´ y pomoc´ı webov´eho prohl´ıˇzeˇce a kter´ y funguje n´asleduj´ıc´ım zp˚ usobem. V prvn´ı f´azi uˇzivatel (l´ekaˇr) zvol´ı klinickou ˇc´ast l´ekaˇrsk´e zpr´avy, aby byly odstranˇeny u ´daje, kter´e v˚ ubec zpracov´av´any b´ yt nemaj´ı. Tato ˇc´ast je automatizovanˇe tokenizov´ana a pˇripravena pro druhou f´azi zpracov´an´ı. C´ılem druh´e f´aze je standardizovat vstupn´ı l´ekaˇrskou zpr´avu. Obsahuje nˇekolik n´astroj˚ u pro u ´pravu textu. V´ ysledn´a standardizovan´a zpr´ava by nemˇela obsahovat zkr´acen´a slova, m˚ uˇze obsahovat v klinick´e praxi bˇeˇznˇe uˇz´ıvan´e zkratky, a nesm´ı obsahovat ani pˇreklepy ani jin´e druhy chyb pˇri psan´ı. Ukonˇcen´ı druh´e f´aze umoˇzn´ı uˇzivateli pokraˇcovat do tˇret´ı f´aze zpracov´an´ı. C´ılem tˇret´ı f´aze je identifikace a oznaˇcen´ı slovn´ıkov´ ych term´ın˚ u. Uˇzivatel vyhled´av´a a oznaˇcuje nalezen´e term´ıny identifik´atory jednotliv´ ych podporovan´ ych klasifikaˇcn´ıch syst´em˚ u. Podporovan´ ymi klasifikaˇcn´ımi syst´emy jsou SNOMED CT, LOINC, MKN10 ´ a datab´aze l´ek˚ u a l´eˇciv´ ych pˇr´ıpravk˚ u SUKL.
Podˇ ekov´ an´ı Pr´ace byla ˇc´asteˇcnˇe podpoˇrena projektem SVV-2015260158 Univerzity Karlovy v Praze.
Kl´ıˇ cov´ a slova Zpr´ ava
Strukturovanou informaci z´ıskanou z l´ekaˇrsk´ ych zpr´av lze uloˇzit v elektronick´em zdravotn´ım z´ aznamu a d´ale Definice: Podrobn´ y v´ yˇcet nebo tvrzen´ı nebo form´aln´ı ji vyuˇz´ıt pro rozhodov´ an´ı pˇri poskytov´ an´ı p´eˇce. Strukz´aznam dat poˇr´ızen´ y na z´akladˇe skuteˇcnosti. turovan´a informace by mˇela pˇredstavovat nˇejak´ y druh uˇziteˇcn´eho extraktu, jako napˇr. pacientsk´eho souhrnu ep- Zdroj: http://www.nlm.nih.gov/cgi/mesh/2015/MB_ SOS [4]. Z´akladn´ım oˇcek´ avan´ ym uˇziteˇcn´ ym u ´ˇcelem je pocgi skytnut´ı pˇr´ıstupu k u ´daj˚ um napˇr. o pacientovˇe anamn´eze, preskripci pro poskytov´ an´ı pˇreshraniˇcn´ı p´eˇce. Dalˇs´ım SNOMED CT: 229059009 S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
65
66
Zv´ara K. a kol. – Pˇredzpracov´an´ı l´ekaˇrsk´ych zpr´av pro extrakci informac´ı
MeSH: D058028 ICD10: nenalezeno
Legislativa
Zdroj: http://www.nlm.nih.gov/cgi/mesh/2015/MB_ cgi SNOMED CT: 86078004
MeSH: D011787 Definice: Pr´ace sest´avaj´ıc´ı z textu navrˇzen´e ˇci platn´e legislativy ve formˇe v´ ynos˚ u, z´ akon˚ u, vyhl´aˇsek, ICD10: nenalezeno naˇr´ızen´ı ˇci dalˇs´ıch druh˚ u pr´ avn´ıch pˇredpis˚ u. Zdroj: http://www.nlm.nih.gov/cgi/mesh/2015/MB_ cgi SNOMED CT: nenalezeno MeSH: D020485 ICD10: nenalezeno
Kvalita p´ eˇ ce ´ Definice: Uroveˇ n, kter´ a charakterizuje zdravotnictv´ı nebo poskytovan´e zdravotn´ı sluˇzby zaloˇzen´a na pˇrijat´ ych standardech kvality.
Reference [1] Garcia-Remesal M., Maojo V., Billhardt H., Crespo J., Integration of Relational and Textual Biomedical Sources, Methods Inf Med 2009;48(1):76-83 [2] Tsung O. Cheng, Letters to Editor; in: Medical Abbreviations Journal of the Royal Society of Medicine, 97 (11), 2004: 556 [3] Semeck´ y J., Zv´ arov´ a J.(supervisor), Multimedia electronic health record in cardiology. Diploma thesis, Faculty of Mathematics and Physics of Charles University in Prague, 2001 (in Czech) [4] Smart Open Services for European Patients, D3.2.2 Final definition of functional service requirements – Patient Summary, www.epsos.eu (last access 14.7.2015)
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
ˇ Zivicov´ a V. a kol. – Stroma dlaˇzdicobunˇeˇcn´ych n´ador˚ u hlavy a krku
Stroma dlaˇ zdicobunˇ eˇ cn´ ych n´ ador˚ u hlavy a krku ˇ Veronika Zivicov´ a1,2 , Zdenˇ ek F´ık1,2 , Barbora Dvoˇr´ ankov´ a2 , Karel Smetana Jr.2 1 2
Anatomick´y u ´stav 1.LF UK, Prague, Czech Republic
Klinika ORL a chirurgie hlavy a krku 1.LF UK a FN Motol, Prague, Czech Republic
Kontakt: ˇ Veronika Zivicov´ a Anatomick´ yu ´stav 1.LF UK, Prague Adresa: U Nemocnice 3, 128 08 Prague 2 E–mail:
[email protected]
C´ıle v´ yzkumu
saminoglykany jsou v souˇcasnosti detailnˇe zkoum´any pro svou roli v progresi n´adorov´ ych onemocnˇen´ı a jako objekty moˇzn´e c´ılen´e l´eˇcby [6, 1]. Pˇr´ıtomnost tenascinu a jeho vliv na chov´an´ı n´adoru se hodnot´ı i v naˇs´ı laboratoˇri. Tento projekt studuje stroma za in vivo a in vitro podm´ınek a porovn´av´a je s klinick´ ym pr˚ ubˇehem. Vzorky HNSCC byly odebr´any peroperaˇcnˇe na ORL klinice na z´akladˇe informovan´eho souhlasu od 54 pacient˚ u. Byly zpracov´any technikou zmrazen´ ych ˇrez˚ u s n´aslednou pˇr´ıpravou histologick´eho prepar´atu. Tenascin byl detekov´an v t´emˇeˇr 80% pˇr´ıpad˚ u. Jako kontrola byla pouˇzita zdrav´a buk´aln´ı sliznice. Dle naˇsich v´ ysledk˚ u se tenascin nach´az´ı ve zdrav´e tk´ani pouze v baz´aln´ı vrstvˇe, zat´ımco v n´adorech je v cel´e ˇs´ıˇri epitelu. Na z´akladˇe dispenzarizace pacient˚ u byla vytvoˇrena Kaplan-Meierova anal´ yza pˇreˇzit´ı. Anal´ yza uk´azala, ˇze pacienti s pozitivitou tenascinu ˇzij´ı kratˇs´ı dobu. Zjiˇstˇen´a z´avislost vˇsak nebyla statisticky v´ yznamn´a. D´ale byly ze 4 n´adorov´ ych a 3 zdrav´ ych tk´an´ı z´ısk´any prim´arn´ı fibroblasty. Fibroblasty byly kultivov´any jednak za standardn´ıch podm´ınek, jednak za stimulace TGFb-1. V in vitro kultivaci zdrav´ ych fibroblast˚ u za standardn´ıch podm´ınek po dobu 7 dn´ı se tenascin nenach´az´ı. Stimulovan´e kultury jsou tenascin pozitivn´ı. N´adorovˇe asociovan´e fibroblasty jsou naproti tomu heterogenn´ı a na stimulaci neodpov´ıdaj´ı v pozitivn´ım ani negativn´ım smyslu. V produkci extracelul´arn´ı matrix jsou velmi r˚ uznorod´e. Tato heterogenita m˚ uˇze b´ yt podkladem rozd´ıln´eho chov´an´ı n´ador˚ u pozorovan´eho u pacient˚ u. Experimenty potvrzuj´ı d˚ uleˇzitou roli stromatu v biologii n´ador˚ u hlavy a krku. Jeho komponenty jako fibronektin nebo tenascin ovlivˇ nuj´ı r˚ ust a ˇs´ıˇren´ı n´adoru. Z tohoto d˚ uvodu je nutn´e bliˇzˇs´ı pozn´an´ı jeho jednotliv´ ych sloˇzek. Tyto molekuly maj´ı velk´ y klinick´ y v´ yznam, protoˇze mohou slouˇzit jako vhodn´e nov´e c´ıle protin´adorov´e terapie.
Mezi n´adory hlavy a krku dominuj´ı dlaˇzdicobunˇeˇcn´e karcinomy (90%), vych´ azej´ıc´ı ze sliznic horn´ıch d´ ychac´ıch a polykac´ıch cest, nejˇcastˇeji dutiny u ´stn´ı, orofaryngu, hypofaryngu a hrtanu – dlaˇzdicobunˇeˇcn´e karcinomy hlavy a krku (HNSCC – head and neck squamous cell carcinoma) [2, 7]. Podp˚ urnou sloˇzkou malignˇe transformovan´ ych epitel˚ u je stroma sest´ avaj´ıc´ı z extracelul´ arn´ı matrix (ECM), n´adorovˇe asociovan´ ych fibroblast˚ u, myofibroblast˚ u a dalˇs´ıch bunˇek (pericyty, hladk´e svalov´e buˇ nky, adipocyty, makrof´agy, mastocyty, lymfocyty) [9]. Myofibroblast od fibroblastu odliˇsuje pˇr´ıtomnost kontraktiln´ıch mikrofilament. Za spolehlivou kombinaci znak˚ u identifikuj´ıc´ıch myofibroblast jsou povaˇzov´ any hladk´ y svalov´ y aktin (SMA), P4H, vimentin a absence cytokeratin˚ u. Myofibroblasty negativnˇe ovlivˇ nuj´ı pr˚ ubˇeh chronick´eho z´anˇetu. V n´ adorech, kde vznikaj´ı z n´adorovˇe asociovan´ ych fibroblast˚ u (CAFs), indukuj´ı progresi onemocnˇen´ı [12]. Tato podobnost je zahrnuta v tezi Harolda Dvoraka o n´adoru jako r´ anˇe, kter´ a se nehoj´ı [5]. P˚ uvod CAFs nen´ı dosud objasnˇen. Zvaˇzuje se vznik z lok´aln´ıch mezenchymov´ ych bunˇek, z mezenchymov´e kmenov´e buˇ nky kostn´ı dˇrenˇe a do tˇretice z n´ adorov´e buˇ nky cestou epitelo-mezenchymov´e transformace [4]. Dalˇs´ı v´ yznamnou sloˇzkou n´ adorov´eho stromatu je fibronektin, pod´ılej´ıc´ı se na proliferaci fibroblast˚ u, stimulaci chemotaxe imunokompetentn´ıch bunˇek, stimulaci produkce prote´az atd. [11]. V dlaˇzdicobunˇeˇcn´em karcinomu dutiny u ´stn´ı je pˇr´ıtomnost fibronektinu asociov´ana s pˇr´ıtomnost´ı uzlinov´ ych metast´ az a tud´ıˇz celkovou horˇs´ı progn´ozou pro pacienta [8, 3]. V extracelul´arn´ı matrix HNSCC lze detekovat tak´e tenascin. Tenascin zahrnuje rodinu glykoprotein˚ u, ze kter´e je nejl´epe prozkoum´ an tenascin-C. Tenascin-C hraje casn´ y stav pozn´ an´ı d˚ uleˇzitou roli jak ve fyziologick´ ych, tak patologick´ ych pro- Souˇ cesech. Ve zv´ yˇsen´e m´ıˇre byl nalezen nejen v HNSCC, ale tak´e v karcinomech kolorekta, prsu, moˇcov´eho mˇech´ yˇre, Dlaˇzdicobunˇeˇcn´e n´adory hlavy a krku sest´avaj´ı plic, prostaty a v gliomu. Jedna jeho podjednotka (FN III) z n´adorov´ ych bunˇek a n´adorov´eho stromatu. N´adorov´e je schopn´a v´azat fibronektin. [10] Tenascin spolu s glyko- stroma je podp˚ urn´a tk´an ˇ, kter´a obsahuje extracelul´arn´ı S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
67
68
ˇ Zivicov´ a V. a kol. – Stroma dlaˇzdicobunˇeˇcn´ych n´ador˚ u hlavy a krku
matrix, n´adorovˇe asociovan´e fibroblasty a dalˇs´ı bunˇeˇcn´e Tenascin typy [9]. I kdyˇz nov´e l´eˇcebn´e modality m´ıˇr´ı specificky proti n´adorov´e buˇ nce, dlaˇzdicobunˇeˇcn´e n´ adory hlavy a krku Definice: Hexametrick´ y glykoprotein extracelul´arn´ı maˇcasto recidivuj´ı a metastazuj´ı. Pˇr´ıˇcina by mohla b´ yt trix pˇrechodnˇe exprimovan´ y v mnoha vyv´ıjej´ıc´ıch v n´adorov´em stromatu a jeho sloˇzk´ ach, jako je fibronektin se org´anech a ˇcasto re-exprimovan´ y v n´adorech. nebo tenascin [11, 1]. Je pˇr´ıtomn´ y v centr´aln´ım a perifern´ım nervov´em syst´emu, tak jako v hladk´em svalu a ˇslach´ach.
Aplikace v biomedic´ınˇ e a zdravotnictv´ı
Zdroj: nenalezeno SNOMED CT: nenalezeno
Sloˇzky extracelul´arn´ı matrix hraj´ı v´ yznamnou roli MeSH: D019063 v n´adorov´e biologii. Glykoproteiny extracelul´ arn´ı matrix ovlivˇ nuj´ı n´adorov´ y r˚ ust a ˇs´ıˇren´ı. Proto by tyto molekuly ICD10: nenalezeno mohly b´ yt pouˇzity jako nov´e c´ıle protin´ adorov´e terapie.
Reference
Podˇ ekov´ an´ı Pr´ace byla ˇc´asteˇcnˇe podpoˇrena projektem SVV-2015260158 Univerzity Karlovy v Praze.
Kl´ıˇ cov´ a slova Karcinom Definice: Malign´ı epitelov´ a neopl´ azie Zdroj: nenalezeno SNOMED CT: 68453008 MeSH: D002277 ICD10: nenalezeno
Extracelul´ arn´ı matrix Definice: S´ıtovit´a struktura nalezena v extracelul´ arn´ım prostoru a ve spojen´ı s baz´ aln´ı membr´ anou povrchu buˇ nky. Podporuje bunˇeˇcnou proliferaci a podporuje adhezi jak bunˇeˇcn´ ych struktur, tak lyz´ atu v kultuˇre. Zdroj: Kreis & Vale, Guidebook to the Extracellular Matrix and Adhesion Proteins, 1993, p93 SNOMED CT: nenalezeno MeSH: D005109 ICD10: nenalezeno
[1] Afratis N, Gialeli C, Nikitovic D, Tsegenidis T, Karousou E, Theocharis AD, Pavao MS, Tzanakakis GN, Karamanos NK. Glycosaminoglycans: key players in cancer cell biology and treatment. FEBS J. 2012; 279(7):1177-97. [2] Argiris A, Karamouzis MV, Raben D, Ferris RL. Head and neck cancer. Lancet. 2008; 371(9625):1695-709. [3] de Bondt RB, Nelemans PJ, Hofman PA, Casselman JW, Kremer B, van Engelshoven JM, Beets-Tan RG. Detection of lymph node metastases in head and neck cancer: a meta-analysis comparing US, USgFNAC, CT and MR imaging. Eur J Radiol. 2007; 64(2):266-72. [4] De Wever O, Demetter P, Mareel M, Bracke M. Stromal myofibroblasts are drivers of invasive cancer growth. Int J Cancer. 2008; 123(10):2229-38. [5] Dvorak HF. Tumors: wounds that do not heal. Similarities between tumor stroma generation and wound healing. N Engl J Med. 1986; 315(26):1650-9. [6] Guttery DS, Shaw JA, Lloyd K, Pringle JH, Walker RA. Expression of tenascin-C and its isoforms in the breast. Cancer Metastasis Rev. 2010;29(4):595-606. [7] Licitra L, Felip E, Group EGW. Squamous cell carcinoma of the head and neck: ESMO clinical recommendations for diagnosis, treatment and follow-up. Ann Oncol. 2009; 20 Suppl 4:121-2. [8] Lyons AJ, Bateman AC, Spedding A, Primrose JN, Mandel U. Oncofetal fibronectin and oral squamous cell carcinoma. Br J Oral Maxillofac Surg. 2001; 39(6):471-7. [9] Polyak K, Haviv I, Campbell IG. Co-evolution of tumor cells and their microenvironment. Trends Genet. 2009; 25(1):30-8. [10] Pas J, Wyszko E, Rolle K, Rychlewski L, Nowak S, Zukiel R, Barciszewski J. Analysis of structure and function of tenascin-C Int J Biochem Cell Biol. 2006;38(9):1594-602. [11] Ritzenthaler JD, Han S, Roman J. Stimulation of lung carcinoma cell growth by fibronectin-integrin signalling. Mol Biosyst. 2008; 4(12):1160-9. [12] Van Buerden HE, Von den Hoff JW, Torensma R, Maltha JC, Kuijpers-Jagtman AM. Myofibroblasts in palatal wound healing: prospects for the reduction of wound contraction after cleft palate repair. J Dens Res. 2005;84(10):871-80.
S´emantick´a interoperabilita v biomedic´ınˇe a zdravotnictv´ı
Part II – English Semantic Interoperability in Biomedicine and Healthcare
Část II – Anglicky
1
Contents
3–5
The Follow-up Management of Thyroid Disorders During the Pregnancy Bart´ akov´ a J., Jiskra J.
6–9
Safety of Private Data in Big Data and Biomedicine Berger J., Beyr K.
10–12
Role of Single Nucleotide Polymorphisms in the Pathogenesis of High-grade Gliomas Bielnikov´ a H., Buzrla P., Bielnik O., Tomanov´ a R., Urbanovsk´ a I., Hruˇskov´ a L., Dvoˇr´ aˇckov´ a J., Mazura I.
13–14
On Gene Conversion Properties Gergelits V.
15–16
Mutation Detection of Collagen Type I Genes Hruˇskov´ a L.
17–26
Statistical Methods for Constructing Gestational Age-Related Charts for Fetal Size and Pregnancy Dating using Longitudinal Data Hynek M., Long J.D., Stejskal D., Zv´ arov´ a J.
27–30
Significance of Cerebral Folate Deficiency for Development and Progression of Autism ˇarek M. Krsiˇcka D., S´
31–33
Integration of Various Lifestyle and Diabetes Devices Within a Diabetes Self-management Application Muˇzn´y M., Vlas´ akov´ a M., Muˇz´ık J., Arsand E.
34–35
Matching Medical Websites to Medical Guidelines through Clinical Vocabularies Rak D., Sv´ atek V.
36–38
International Communication Protocols for Interoperability in the Czech Republic Seidl L., Hanzl´ıˇcek P.
39–41
Big Data in Hospital Information Systems in the terms of Security Schlenker A., Reimer M.
42–44
Secondary Cataract in Patients after Implantation of Multifocal IOLs Siˇcov´ a K., V´yborn´y P., Paˇsta J.
45–46
DNA in Biomedical Applications Slov´ ak D., Zv´ arov´ a J.
47–49
Unstructured Data in Evidence-based Healthcare Stonov´ a M.
50–52
Conduction System Development in Mouse ˇ nkov´ Saˇ a B., Beneˇs J., Sedmera D.
53–54
Presence of Nasal Microbiota and Their Influence on Development of Chronic Rhinosinusitis ˇ Steffl M., Plz´ ak J.
55–57
Glaucoma Treatment for 1 CZK per Day – Dream or Reality? Vesel´ a Fl´ orov´ a Z., V´yborn´y P., Siˇc´ akov´ a S., Obenberger J.
58–60
Telemonitoring of the Basic Therapeutic Elements of the Diabetes Mellitus Treatment and Their Evaluation Vlas´ akov´ a M., Muˇzn´y M., Muˇz´ık J.
61–63
Advantages of Virtual Patient in Paramedical fields of the Health Care Services Vondruˇskov´ a L.
64–65
Preprocessing of Narrative Medical Reports for Information Extraction Zv´ ara K., Tomeˇckov´ a M., Sv´ atek V., Zv´ arov´ a J.
66–67
Stroma of Head and Neck Squamous Cell Carcinoma ˇ Zivicov´ a V., F´ık Z., Dvoˇr´ ankov´ a B., Smetana Jr. K.
Semantic Interoperability in Biomedicine and Healthcare
Bart´akov´a J., Jiskra J. – The Follow-up Management of Thyroid Disorders During the Pregnancy
The Follow-up Management of Thyroid Disorders During the Pregnancy Jana Bart´ akov´ a1,2 , Jan Jiskra1 1 2
Institute of Biophysics and Informatics, First Faculty of Medicine, Charles University in Prague, Prague, Czech Republic
Third Department of Medicine, General University Hospital and First Faculty of Medicine, Charles University in Prague, Prague, Czech Republic
Correspondence to: Jana Bart´ akov´ a Institute of Biophysics and Informatics, First Faculty of Medicine, Charles University in Prague Address: Salmovsk´ a 478/1, 128 00, Prague 2 E–mail:
[email protected]
Aims of Research Our primary aim is to determine if the thyroid checkups among pregnant women on levothyroxine (LT4) treatment during gestation in Czech Republic are in the line with the recommendation of the recent guidelines from the American Thyroid Association (ATA) 2011 or Endocrine Society (ES) 2012. The secondary aim included evaluation of the possibilities to reduce costs by modification of thyroid laboratory utilization in clinical practice.
State of the Art About 10-15% of pregnant women are positive for autoantibodies to thyroid peroxidase (TPOAb) [1, 2, 3] and up to 5% have elevated thyroid stimulating hormone (TSH) [4, 5]. The negative impact of untreated thyroid dysfunction on fertility, course of pregnancy and postpartum period and development of offspring has been well described [5, 6, 7, 8, 9]. However, up to date studies focusing on an appropriate follow-up management of pregnant women with thyroid disorders in clinical practice are lacking. In 2011 and 2012 two guidelines from the ATA and the ES for the management of thyroid disease in pregnancy were published [10, 11]. Both guidelines contain recommendations for the initiation of LT4 treatment and the frequency of thyroid check-ups once treatment is started. There is a high degree of consistency between the guidelines: thyroid function should be tested by TSH every 4-6 weeks in the first trimester of pregnancy. During second and third trimesters the ES recommends to continue testing every 4-6 weeks while the ATA suggests checking TSH level twice during these period of pregnancy (once during the second and once during the third trimester). Semantic Interoperability in Biomedicine and Healthcare
What’s more, not only guidelines recommendations can slightly vary but also the endocrinology practice can vary. In 2013, clinical members of the ES, the ATA, and the AACE (American Association of Clinical Endocrinologists) were asked to fill in a web based survey consisting of 30 questions that dealt with testing, treatment and modulating factors in the management of hypothyroidism. Among 821 respondents, 67.7% would check thyroid laboratory studies every four weeks during pregnancy, 21.4% every eight weeks, 7.9% every 12 weeks, and 2.9% every 2 weeks [12]. In our study from the years 2004-2014, 188 pregnant women treated with LT4 were followed and examined during their pregnancy according to the recommended algorithm by ATA 2011. The thyroid check-up in third trimester (26th-32th gestational week) was evaluated as redundant in case of uncomplicated hypothyroidism and/or positive TPOAb and producing inadequate high costs versus low likelihood of benefit. As we can see, even the thyroid management during pregnancy is in principle straightforward, it can’t be the case of clinical practice. Therefore, the thyroid laboratory utilization during pregnancy should be more study.
Application in Biomedicine and Healthcare The inappropriate laboratory utilization is widely prevalent and costly for healthcare system [13, 14, 15, 16]. Hence the potentially redundant thyroid tests could be readily modifiable and highly effective component of management of pregnant women with thyroid disorders. This can be made throughout the study of necessity of the recommended thyroid check-up frequency, the real clinical practice of the repeat testing as well as the real investigated hormones in endocrinology practice.
3
4
Bart´akov´a J., Jiskra J. – The Follow-up Management of Thyroid Disorders During the Pregnancy
Acknowledgements
Antibodies to Thyroid Peroxidase
This paper has been partially supported by the SVV- Definition: Antibodies targeted against thyroid peroxidase, an enzyme normally found in the thyroid 2015-260158 project of Charles University in Prague. gland, plays an important role in the production of thyroid hormones.
Keywords
Synonyms: Antithyroperoxidase antibody, Thyroid peroxidase antibody, TPO antibodies, Thyroperoxidase antibodies, anti-thyroid peroxidase antibodies, antiTPO antibodies Definition: Any benign or malignant condition that affects the structure or function of the thyroid gland. Reference: Clinique health [online]. 2015 [cit. 2015-23-06]. Available from: http://www. Synonyms: Disorder of thyroid gland, Disease of thyroid cliquehealth.org/index.php?option=com_ gland, Thyroid disease, Disorder of thyroid zoo&task=item&item_id=1505&Itemid=4 Reference: The Free Dictionary [online]. 2015 [cit. 2015-23-06]. Available from: https:// SNOMED CT: 103184005 medical-dictionary.thefreedictionary.com/ MeSH: not found SNOMED CT: 14304000 ICD10: Y42.8 MeSH: C19.874
Thyroid Disorder
ICD10: E00-E07
Pregnancy Definition: The period from conception to birth.
Levothyroxine Definition: Thyroid hormone prepared synthetically; used as the sodium salt in the treatment of hypothyroidism and the treatment and prophylaxis of goiter and thyroid carcinoma.
Synonyms: Gestation, Gestational period, Pregnant Synonyms: Levothyroxine sodium, Thyroxine sodium, State Thyroxine, L-Thyroxine, L-Thyrox, Levo-T, Sodium Levothyroxine Reference: The Free Dictionary [online]. 2015 [cit. 2015-23-06]. Available from: https:// Reference: The Free Dictionary [online]. 2015 medical-dictionary.thefreedictionary.com/ [cit. 2015-23-06]. Available from: https:// medical-dictionary.thefreedictionary.com/ SNOMED CT: 289908002 SNOMED CT: 126202002 MeSH: G08.686.785.769 MeSH: D06.472.931.812; D12.125.072.050.767 ICD10: Z33 ICD10: Y42.1
Thyroid Stimulatory Hormone
References Definition: A hormone that is secreted by the anterior lobe of the pituitary gland and stimulates the thyroid gland. Synonyms: Thyroid Stimulating Hormone, Thyroid Stimulation Hormone, TSH, Thyrotrophin, Thyrotropin, Thyrotropic hormone Reference: Merriam Webster: An Encyclopedia Britannica Company [online]. 2015 [cit. 2015-23-06]. Available from: http://www.merriam-webster. com/ SNOMED CT: 65428006 MeSH: D06.427.699.631.525.883; D12.644.548.691.525.883 ICD10: Y42.8
[1] Springer D, Zima T, Limanova Z. Reference intervals in evaluation of maternal thyroid function during the first trimester of pregnancy. European journal of endocrinology / European Federation of Endocrine Societies. 2009;160(5):791-7. [2] Lazarus JH, Kokandi A. Thyroid disease in relation to pregnancy: a decade of change. Clinical endocrinology. 2000;53(3):265-78. [3] Glinoer D. The regulation of thyroid function in pregnancy: pathways of endocrine adaptation from physiology to pathology. Endocrine reviews. 1997;18(3):404-33. [4] Potlukova E, Potluka O, Jiskra J, Limanova Z, Telicka Z, Bartakova J, et al. Is age a risk factor for hypothyroidism in pregnancy? An analysis of 5223 pregnant women. The Journal of clinical endocrinology and metabolism. 2012;97(6):1945-52. [5] Allan WC, Haddow JE, Palomaki GE, Williams JR, Mitchell ML, Hermos RJ, et al. Maternal thyroid deficiency and pregnancy complications: implications for population screening. Journal of medical screening. 2000;7(3):127-30.
Semantic Interoperability in Biomedicine and Healthcare
Bart´akov´a J., Jiskra J. – The Follow-up Management of Thyroid Disorders During the Pregnancy
[6] Lazarus JH. Thyroid function in pregnancy. British medical bulletin. 2011;97:137-48. [7] Krassas GE, Poppe K, Glinoer D. Thyroid function and human reproductive health. Endocrine reviews. 2010;31(5):702-55. [8] Casey BM, Dashe JS, Wells CE, McIntire DD, Leveno KJ, Cunningham FG. Subclinical hyperthyroidism and pregnancy outcomes. Obstetrics and gynecology. 2006;107(2 Pt 1):337-41. [9] Benhadi N, Wiersinga WM, Reitsma JB, Vrijkotte TG, Bonsel GJ. Higher maternal TSH levels in pregnancy are associated with increased risk for miscarriage, fetal or neonatal death. European journal of endocrinology / European Federation of Endocrine Societies. 2009;160(6):985-91. [10] Stagnaro-Green A, Abalovich M, Alexander E, Azizi F, Mestman J, Negro R, et al. Guidelines of the American Thyroid Association for the diagnosis and management of thyroid disease during pregnancy and postpartum. Thyroid : official journal of the American Thyroid Association. 2011;21(10):1081-125. [11] Lazarus J, Brown RS, Daumerie C, Hubalewska-Dydejczyk A, Negro R, Vaidya B. 2014 European thyroid association guide-
Semantic Interoperability in Biomedicine and Healthcare
lines for the management of subclinical hypothyroidism in pregnancy and in children. European thyroid journal. 2014;3(2):7694. [12] Burch HB, Burman KD, Cooper DS, Hennessey JV. A 2013 survey of clinical practice patterns in the management of primary hypothyroidism. The Journal of clinical endocrinology and metabolism. 2014;99(6):2077-85. [13] Leese B. Is there too much laboratory testing? Reprt 79. York, Great Britain: University of York. 1991:29pp. [14] Beck JR. Does feedback reduce inappropriate test ordering? Archives of pathology & laboratory medicine. 1993;117(1):334. [15] Bareford D, Hayling A. Inappropriate use of laboratory services: long term combined approach to modify request patterns. BMJ (Clinical research ed). 1990;301(6764):1305-7. [16] van Walraven C, Raymond M. Population-based study of repeat laboratory testing. Clinical chemistry. 2003;49(12):19972005.
5
6
Berger J., Beyr K. – Safety of Private Data in Big Data and Biomedicine
Safety of Private Data in Big Data and Biomedicine Jiˇr´ı Berger1 , Karel Beyr 1
Institute of Pathological Physiology, Prague, Czech Republic
Correspondence to: Jiˇr´ı Berger Institute of Pathological Physiology Address: U Nemocnice 5, 128 53 Prague 2 E–mail:
[email protected]
Aims of Research
data set as possible, the higher benefit will they have for future processing of various analyses based on health docBig Data have a great potential for research in umentation of whole population and concerning biomedibiomedicine in many areas cal information. • Analysis of patient segmentation, treatment price Once such a project starts – and as authors of this artiand result helps determine the medically and ecocle assume, it is not going to be a technological, but rather nomically most efficient course of treatment for a an organizationally ethical problem – the most effective given patient means of processing such a large amount of data will be • Proactive identification of patients who would bene- to provide it to the professional public as a source of refit from preventive health care search. It is quite common abroad when a project is fully or even only partially funded from public resources then • Analysis of disease incidence can provide epidemiothe conditions are set so that the information is available logical findings and suggest preventive measures for non-commercial activities with minimal limitations. In • Assisting in detecting and minimizing fraud at- spite of the fact that it will never be possible to fully disclose biomedical information (with regard to sensitivity of tempts in health care stored data), there still exists a large array of uses, im• Cooperation with pharmaceutical companies, so plementations and applications that would benefit from that it will be easier for them to identify group of having access to it. relevant patients for clinical trials (assuming the patients’ prior consent) Due to the use of population data, there exists a risk of Nowadays, the trend of digitalization of medical and indirect identification of patients’ information. The slightrelated documents marks a time for engagement of tech- est sign of abuse brings ethical problems and may even nology called Big Data for biomedical informatics. This stop the entire research. technology provides faster and more efficient processing and sharing of huge amount of data. Since health care inTo prevent possible issues with privacy, it is necessary volves sensitive data, the main concern is a protection of to enforce very strong and efficient rules that facilitate patients’ private data. Many countries are implementing maximum data yield, together with a strict conservation computerization of health care. For example, in the USA of anonymity and protection of private data. It is necesa ”Health Information Technology for Economic and Clinsary to limit data mining so that it would not be possible ical Health Act”, (HITECH) is being employed. The aim to abuse or disclose sensitive data by any, even theoretical, of this research is to design and define the rules that premeans. Abroad, a relevant legislative is already in place, vent the abuse and fraud concerning sensitive biomedical e.g. the current version of ”Health Insurance Portability data, but which would not limit its efficiency and quality and Accountability Act” (HIPAA) in the USA specifies of output data at the same time. the standards concerning healthcare records transactions. Similarly, the EU legislation called Data Protection DiState of the Art rective 95/46/EC defines the necessity of patient consent concerning processing of his private data and portability The larger the quantity of heterogeneous biomedical of medical data. However, EU still does not have a unified data grouped in Big Data so that they contain as complex approach to the protection of private data [1]. Semantic Interoperability in Biomedicine and Healthcare
Berger J., Beyr K. – Safety of Private Data in Big Data and Biomedicine
Application in Biomedicine and Healthcare
(name, national identification number) were replaced, it is necessary to manage access restrictions. In such cases, it is often possible to gain specific patient data, or at least data which can be inferred with a high degree of probability. For aforementioned reasons, it is necessary to employ a solution that can limit queries, combination of which can reveal sensitive data, or only enable those combinations to personnel with higher access rights while ensuring feedback control and risk analysis of searched queries and their results. Another query which can lead to sensitive data breach is the one that returns significantly small set of result entities. Queries can contain combination of various factors. However, if a query is ”over combined”, it can, in extreme case, lead to a scenario in which only one patient is in the result set. Even if the patient’s name is not stored in database, it can sometimes be inferred. For example, if we know a fraction of health record of a given person, then using relevant information (age, sex and address) we can indirectly gain his sensitive information. This risk can be eliminated to some extent by employing heuristic rules and their gradual improvement. Those rules would block answers that could contain risky set of information.
The usage of Big Data in biomedicine and healthcare will always have its specifics. The amount of anonymization used on data will always be inversely proportional to the quality of output data [2]. It means, that one of the key elements of successfully using Big Data will be setting of the ratio between anonymization and the quality of mined data. Basic anonymization would be required for efficient usage. By basic anonymization we consider removal of (or in any way denying access to) private information like name and national identification number and their replacement with anonymous identifier which would identify subject across the data set. Unfortunately, data altered in such a way will still be vulnerable. Therefore it is necessary to prevent various types of possible privacy attacks. For example, a query returning prescribed medicaments and their dosage for concrete patient contains sensitive data. From the knowledge of medicine prescribed, it is possible to infer patient diagnosis. If the private data (name, national identification number) are anonymized, then we can assume that returned data will not contain sensitive information. On the other hand there exists a plethora of queries that do not return sensitive data. For example: Query about the amount of practitioner’s patients, query about Indirect disclosure of sensitive data from fully prescribed medicine in certain region, or query about spe- anonymized database cific diagnosis across the population. In case of anonymized data set, full access to database can be provided under specified conditions. Securing the whole database Before the medical records are saved to a database, it is possible (or sometimes even required by relevant legislaOne possibility to increase security in biomedicine and tion) to anonymize the records (remove name and national health care is to encrypt the underlying data. It adds anidentification number) and also generalize them. We call other safety layer and thus decreases the risk of sensitive this process full anonymization. By generalization we data leak or abuse [3]. mean making identification a person by quasi-identifiers There are various advanced algorithms [4] which can (eg. date of birth, address, sex) difficult or impossible. encrypt medical records so that only personnel with relBy matching quasi-identifiers, medical records the goverevant authorization can decode them. Those algorithms nor of Massachusetts were leaked using quasi-identifiers have advantages over classical encrypting methods (symthat were accessible in anonymized mode and matched metrical and asymmetrical ciphers) – they are faster and with electoral data containing quasi-identifiers that were cheaper than traditional RSA concept, and they provide published together with names. better security in case of stolen password. Authors of [5] Author of [6] describe generalization algorithm as a describe querying of medical data encrypted using these principle that patients form groups based on their quasialgorithms. Their findings are perfectly suitable for Big identifiers, and each group must contain at least K paData concept. tients. As encryption will inevitably bring slower querying, it This approach makes it difficult to identify people, but should not be recommended for the whole data set, but in some cases identification is still possible in contrary to only for structured patient data. K-anonymity. Authors of [7] improve K-anonymization by proposing improvement, L-anonymization. It requires Indirect disclosure of sensitive data from patients in the same quasi-identifier group to have heteropartially anonymized database geneous sensitive data. Another approach to generalization lies in rounding If there are completely non-anonymized data, or data quasi-identifiers. Data can be stored in database in more with basic anonymization in which the base patient data forms, each time with different level of precision. The Semantic Interoperability in Biomedicine and Healthcare
7
8
Berger J., Beyr K. – Safety of Private Data in Big Data and Biomedicine
higher authorization level of the personnel reading data, the more precise quasi-identifiers can be accessed. For example, instead of a birth date, only a birth year or a decade is stored. Instead of a whole address, only a city or region name is stored. Another possibility of generalization lies in not storing certain quasi-identifiers at all. ICD codes (International Statistical Classification of Diseases and Related Health Problems), are used abroad. They are maintained by WHO. These codes have hierarchical structure, therefore they are perfectly suitable for generalization. Authors of [8] specify probability of patient identification based on frequency of rare ICD codes. They recommend a removal of 5% to 25% of the rarest codes, and their replacement by more generic ones. This leads to significantly lower probability of patient identification with relatively low precision lost.
Non-triviality of querying A big problem concerning research on Big Data in biomedicine informatics concerns the creation of queries. It is not expected that a majority of researches in biomedical field would be able and willing to design their own Map/Reduce parallel algorithms to solve queries in Big Data medical database. It is more probable that there will be a cooperation from IT technicians, analysts and programmers who will create tools that could be parametrized, run on demand etc. Both issues (safety and non-triviality) could be solved by a query tool. The tool would contain query ”templates”, programs and algorithms that could be parametrized via user interface. The Big Data database could then be simply queried by such a tool without complex training. We advocate using templates primarily as a means of simplification for broad research community and to increase availability of relevant research and its results for wide range of applications. The question of security would also be settled. Template examples: • Prescription of concrete active substance according to medical specialization • Frequency analysis according to place of residence • Demographic composition of patients • Volumes of medical actions of a facility by period • Correlation between diseases and patient’s type of profession Multilayer architecture can be built for this need. It will be an extension of classical Big Data technologies in specific biomedical implementation. It will be split at least into those layers:
1. Professional public will be granted permission to use only prepared templates into which it would be possible to insert custom parameters, but it would not be possible to change the nature of a query. Query could utilize only one template, it would not be possible to combine templates. This way it is guaranteed that no sensitive data could leak. This access will be primarily for postgraduate students for their basic research. 2. Specialized workplaces would be allowed to combine templates and will have greater freedom in their parametrization. A set of heuristic rules will oversee the queries and will report or even block the combinations that could lead to a disclosure of sensitive information. 3. Team of analysts with security credentials will prepare templates including heuristic rules that will watch over their usage. Under standard security checks they could also perform Big Data queries. This practice could be used in cases where there will be a probability of work with sensitive data. As part of their workload, they could handle complex tasks and queries according to requests of individual workplaces in cases when it would not be efficient to use classic templates or when there would be a risk of leak of sensitive data. Resulting data would be checked and possibly anonymized before their return to requesting workplace. 4. Narrow specialized team of auditors will define and configure advanced heuristic rules and approve templates before release. 5. The last level could be based on a system with elements of artificial intelligence. It would be based on advanced pattern recognition algorithms, neural networks and learning process. It could automatically scan and detect unhandled possibilities of data abuse in real time.
Discussion Research will continue in three main fields. Firstly, it will be aimed on standard safety rules and relevant security technologies and their utilization in biomedical data. Their specifics and deviations from standard approach in data security methods will be defined. This part will concentrate specifically on analysis of encryption algorithms with regard to granted permissions, their benefits and disadvantages and also their influence on efficient data analysis. Second research area will focus on anonymization algorithms and their influence on data yield and efficiency of processing. It will define the principles and the influence of anonymization on biomedical data with regard to theoretical possibilities of retrieving sensitive data using Semantic Interoperability in Biomedicine and Healthcare
Berger J., Beyr K. – Safety of Private Data in Big Data and Biomedicine
various query combinations. This area will not only be focused on theoretical field, but it will also try to generalize results received by combinatorial methods from representative sample of anonymized data and to compare them with real data and evaluate their similarity. The third and largest area is to define an interface between low-level Big Data querying mechanism and query templates, where the main objective will be a balance of safety mechanisms and usability for wide professional public. Target state would be to find an interface which would be comparable in usability to MS Excel or MS Access, or their alternatives. This area would elaborate on distribution of rights and responsibilities among four defined roles, their detailed description and process mapping of their relation to data security. Each role will be analyzed for potential risks and threats including relevant countermeasures. Basic analysis of heuristic functions and elements of artificial intelligence as means of improving safety of biomedical data will be performed.
Acknowledgements
large hidden values from large datasets that are diverse, complex, and of a massive scale. Synonyms: Computer cloud, multi-node database Reference: [6] SNOMED CT: not found MeSH: not found ICD10: not found
Anonymization Definition: Process of either encrypting or removing personally identifiable information from data sets, so that the people whom the data describe remain anonymous. Synonyms: Information sanitization Reference: [6]
SNOMED CT: not found This paper has been partially supported by the SVVMeSH: not found 2015-260158 project of Charles University in Prague. ICD10: not found
Keywords References Personal Data Definition: Any information relating to an identified or identifiable natural person. Synonyms: Personally identifiable information Reference: art. 2 let. a) Directive n. 95/46/ES on the protection of individuals with regard to the processing of personal data and on the free movement of such data
[1] Boussi Rahmouni H, Solomonides T, Casassa Mont M, Shiu S. Modelling and Enforcing Privacy for Medical Data Disclosure across Europe. In Adlassnig KP, editor. Medical Informatics in a United and Healthy Europe – Proceedings of. Sarajevo: IOS Press; 2009. p. 695-699. [2] Duncan et al. Disclosure Risk vs. Data Utility: The R-U Confidentiality map: Los Alamos National Library; 2001. [3] Amazon Web Services. Creating Healthcare Data Applications to Promote HIPAA and HITECH Compliance. 2012.
SNOMED CT: not found
[4] Alshehri, Radziszowski, Raj. Designing a Secure Cloud-Based EHR System using Ciphertext-Policy Attribute-Based Encryption.
MeSH: not found
[5] Narayan S, Gagn´ e M, Reihaneh SN. Privacy preserving EHR system using attribute-based infrastructure.
ICD10: not found
[6] Sweeney L. k-anonymity: a model for protecting privacy. International Journal on Uncertainty. 2002; 10(5): p. 557-570.
Big Data
[7] Machanavajjhala A, Kifer D, Gehrke J, Venkitasubramaniam M. L-diversity: Privacy beyond k-anonymity. ACM Transactions on Knowledge Discovery from Data. 2007 March; 1(1).
Definition: Big data is a set of techniques and technologies that require new forms of integration to uncover
[8] Vinterbo S, L OM, S D. Hiding information by cell suppression. In Proc AMIA Symp; 2001. p. 726–730.
Semantic Interoperability in Biomedicine and Healthcare
9
10
Bielnikov´a H. et al. – Role of Single Nucleotide Polymorphisms in the Pathogenesis of High-grade Gliomas
Role of Single Nucleotide Polymorphisms in the Pathogenesis of High-grade Gliomas Hana Bielnikov´ a1,2 , Petr Buzrla1 , Ondˇrej Bielnik3 , Radoslava Tomanov´ a1 , Irena Urbanovsk´ a4 , Lucie Hruˇskov´ a2 , Jana Dvoˇr´ aˇ ckov´ a1 , Ivan Mazura2 1
The Institute of pathology, University Hospital Ostrava, Ostrava, Czech Republic 2
First Faculty of Medicine Charles University Prague, Prague, Czech Republic 3
Neurosurgery clinic, University Hospital Ostrava, Ostrava, Czech Republic 4
CGB laboratory Ostrava a.s., Ostrava, Czech Republic
Correspondence to: Hana Bielnikov´ a The Institute of pathology, University Hospital Ostrava Address: 17. listopadu 1790, 70852, Ostrava-Poruba E–mail:
[email protected]
Aims of Research
The gold standard for the diagnosis of gliomas is histological examination, which, however, is complicated by significant heterogeneity of tumors and overlapping morphological characteristics, particularly at higher degrees. For this reason, in recent years, we have been using findings from the field of genetics in the diagnosis and treatment of gliomas. Using cytogenetic and molecular-genetic methods have been discovered a number of alterations of chromosomes and genes which play a significant role in the typing of tumors or are important predictive and prognostic factors [4].
Single nucleotide polymorphisms (SNPs) are single nucleotide variations in the sequence of deoxyribonucleic acid (DNA) between two individuals. This is one of the most common genetic changes in coding and noncoding regions of human genome [1]. The effects of these changes are reflected depending on the location of polymorphisms in DNA. If SNPs located in noncoding regions, their influence on the organism is usually negligible and usually does not lead to phenotypic abnormalities. In the presence of polymorphisms in coding regions, their potential A number of genetic changes, which occur in gliomas, impact on the phenotype is determined by a number of are very broad. These modifications both at the level conditions. of whole chromosomes and their parts (aneuploidy, deleThe aim of my work is the isolation of DNA from tions, duplications, translocations, etc.) and at the level tumor tissue of patients with diagnosed astrocytoma or of DNA strand (point mutations, deletions, insertions, glioblastoma and subsequent analysis of selected genes us- amplifications, substitutions, epigenetic changes- methying molecular-genetic methods. The obtained phenotypic, lation, acetylation, etc.). Most of these alterations conclinical and genetic data will be assessed for their possible cerns areas in which there are important tumor suppressor genes or protooncogens [5, 6]. correlation. In recent years, many studies have been focused on the connections between genetic polymorphisms and the deState of the Art velopment of cancer. One of the discussed polymorphism, also in gliomas, is SNP309 (rs2279744) in the promoter Gliomas are the most common primary brain tumors region of MDM2. In this case polymorphism SNP309 in adults. The prognosis of these tumors, depending on increased binding affinity of the transcriptional activator their degree, is very poor. In malignant types such as Sp1 to the DNA and thereby increases the expression of anaplastic astrocytoma (AA, grade III.) and glioblastoma the MDM2 gene [7]. MDM2 gene product is a protein (GBM, Grade IV.) the time of survival is in the range of regulating p53 pathway, one of the most important tumor 3 to 5 years for AA and 15 to 16 months in GBM [2]. suppressor factors. MDM2 protein as a controller directly The risk of these tumors is their infiltrative growth in the binds to p53 and adversely affects its stability and activity. brain tissue impeding their surgical removal, high prolif- It has been shown that even a slight change in the level erative activity, nuclear atypia, presence of angiogenesis of MDM2 significantly influences the suppressor function of p53 and for some types of tumors overexpression of the and necrosis [3]. Semantic Interoperability in Biomedicine and Healthcare
Bielnikov´a H. et al. – Role of Single Nucleotide Polymorphisms in the Pathogenesis of High-grade Gliomas
gene is associated with progression of disease and a weaker response to therapy [8]. SNPs in coding regions of DNA may also act negatively. In some situations a base change in the codon may cause exchange of one amino acid for another. A negative consequence of these substitutions may be a defective or non-functional protein, or the formation of stop codon, the result is the creation of a truncated protein product, often dysfunctional.
Application in Biomedicine and Healthcare
11
Synonyms: Glial cells tumor Reference: http://lekarske.slovniky.cz/pojem/ gliom SNOMED CT: 67271001 MeSH: D005910 ICD10: not found
Single Nucleotide Polymorphism
Definition: Polymorphism of DNA sequence given by variability of only in one base- substitution in one nucleotide of DNA, this substitution must be exMost of polymorphisms have not a direct impact on tended in the population. human health. The localization of these changes within the nucleic acid sequence is substantial. In some cases Synonyms: Polymorphism, SNP their presence can alter the effectiveness of drugs, affect the response of the organism to exposure to harmful sub- Reference: http://www.genomia.cz/cz/slovnik-pojmu/ stances or be associated with the emergence and develSNOMED CT: not found opment of certain diseases. Tracking the connection between polymorphisms and gliomas may bring new knowl- MeSH: D020641 edge about the origin and development of these tumors, where some of the investigated SNPs can serve as poten- ICD10: not found tial prognostic and diagnostic indicators.
Genetic Mutation
Acknowledgements
Definition: Changing heritable genetic information in the DNA related to genes or whole chromosome. The present study was supported by the project SVV According to the site which is affected it may or 260158 of Charles University in Prague. may not affect the function of cells and organisms. Arises spontaneously, or are caused by external factors - mutagens of chemical, physical or biological Keywords factors.
Deoxyribonucleic Acid
Synonyms: Mutation
Definition: A double-stranded polynucleotide consist- Reference: http://lekarske.slovniky.cz/pojem/ mutace ing of two single chains deoxyribonucleotide units. Serves as a carrier of genetic information. SNOMED CT: 55446002 Synonyms: DNA MeSH: D009154 Reference: Alberts B, et al. Essential Cell Biology. EsICD10: not found pero Publishing 1998. SNOMED CT: 24851008 MeSH: D004247 ICD10: not found
Glioma Definition: Tumor of CNS arising from nerve tissue supporting neuroglia, glia. These include astrocytoma, glioblastoma, etc. Gliomas grow in different parts of the brain. They cause epilepsy, mental changes, increase in intracranial pressure, headaches and impaired vision, bearing signs, etc. They exist in malignant and benign variant. Semantic Interoperability in Biomedicine and Healthcare
Phenotype Definition: The observable appearance or property of an individual, as a result of his genotype hereditary dispositions and environmental effects. Synonyms: Reference: http://lekarske.slovniky.cz/pojem/ fenotyp SNOMED CT: 363778006 MeSH: D010641 ICD10: not found
12
Bielnikov´a H. et al. – Role of Single Nucleotide Polymorphisms in the Pathogenesis of High-grade Gliomas
References
[5] Hersh DS, Mehta RI, et al.The Molecular Pathology of Primary Brain Tumors. Path C Rev. 2013 Sept;18(5):210-220.
[1] Clark DP., Molecular biology: Understanding the Genetic Revolution. Elsevier Academic Press, 2005.
[6] Duncan ChG, Yan H. Genomic alterations and the pathogenesis of glioblastoma. Cell Cycle. 2011 Apr;10(8):1174-1175.
[2] Karajannis MA, Zagzag D., editors. Molecular pathology of nervous system tumors. New York, Springer, 2015.
[7] Wan Y, Wu W, Yin Z, Guan P, Zhou B. MDM2 SNP309, genegene interaction, and tumor susceptibility: an updated metaanalysis. BMC Can.2011;11(208):1-9.
[3] Iacob G, Dinca EB. Current data and strategy in glioblastoma multiforme. J Med Lif. 2009 Agu;2(4):386-393. [4] Nikiforova MN, Hamilton RL. Molecular diagnostics of gliomas. Arch Pathol Lab Med. 2011 May;135:558-568.
[8] Bond GL, Levine AJ. A single nucleotide polymorphism in the p53 pathway interacts with gender, environmental stresses and tumor genetics to influence cancer in humans. Onc 2007;26:1317-1323.
Semantic Interoperability in Biomedicine and Healthcare
13
Gergelits V. – On Gene Conversion Properties
On Gene Conversion Properties V´ aclav Gergelits1,2 1
Institute of Molecular Genetics of the AS CR, Prague, Czech Republic
2
First Faculty of Medicine, Charles University, Prague, Czech Republic
Correspondence to: V´ aclav Gergelits Institute of Molecular Genetics of the AS CR, v.v.i. Address: V´ıdeˇ nsk´ a 1083, 142 20 Prague 4 E–mail:
[email protected]
Aims of Research
State of the Art
Meiotic recombination of homologous chromosomes is essential for increasing the intraspecious and interspecious genetic diversity and for the proper segregation of chromosomes to gametes. The process of meiotic recombination starts in the meiotic prophase I, when enzyme SPO11 programatically forms DNA double-strand breaks (DSBs). On one hand the DSB formation is essential for the genetic diversity enhancement, on the other hand all the DSBs have to be repaired for the cell survival. The repair of a particular DSB starts by a release of one strand of DNA, which searches for a sufficiently similar segment in a chromatid of the homologous chromosome. Alternatively, the DSB could be repaired by a sister chromatid. However, in mammals this possibility is only hypothetical on autosomes so far. The recombination and DSB repair occurs after succesfull localization of a homologous segment. There are two possible results of the recombination. The chromatid arms are either exchanged – crossover (CO), or not exchanged – noncrossover (NCO). In both cases, gene conversion, a one-directional transfer of genetic information from the donor chromatid to the acceptor one, occurs. The NCO is generally less understood than CO. The detection of gene conversion is a good way how to show the places in genome where the NCO have occured. This could lead to better understanding of mechanisms of NCO and gene conversion. Research interest in gene conversion is motivated by its roles in evolution and genome dynamics, human pathogenesis as well as its value in exploring mechanistic models of recombination. [1] In my doctoral project I am interested in possibilities of detection of NCO gene conversion events. To this aim the mouse model is used. The aim is to detect and characterize elements of gene conversion in a whole-chromosome scale and subsequently infer the mechanisms by which the gene conversion occurs. The subject of this paper is to summarize the basic properties of gene conversion and NCO in human and in mammals as well as to show their relevance to human diseases.
The sites of CO do not occur randomly in genome. They rather cluster in recombination hotspots, sites where their occurence is much more probable. The length of those sites is on order of kilobases. In mammals, the recombination hotspots are determined almost exlusively by Prdm9 gene. Every allelic form of Prdm9 determines its binding sites – the sites in genome, where a DSB can occur. Recently, it has been shown that NCOs take place in the same hotspots as COs [2]. A variety of studies has provided estimates of gene conversion lengths based on different types of data. The segments of gene conversion occuring during NCO are generally shorter than those occuring during CO. The NCO gene conversion length was estimated to span 55-290 bp [5]. Recently, two studies referred a detection of NCO gene conversion with tract lengths 86 ± 49 bp [4] and 100-1000 bp [2]. The CO gene conversion tracts were directly detected for the first time by tetrad analysis and its length was referred as 626 ± 319 bp and 566 ± 277 bp, respectively for two different hotspots [4]. In spite of the importance of NCOs, the frequency at which they occur in mammals remains widely uncharacterized. Number of DSB occuring in meiosis suggests that NCO occurs in 90% of cases, while CO occurs only in 10% of cases. However, the ratio between NCO and CO widely varies in different hotspots. Ratios as different as NCO:CO 15:1 and 1:1 were observed in two different hotspots studied by Cole and colleagues [4]. The mechanism of conversion could be inferred by its quantitative properties. The classical Szostak’s model [6] assumed that in mammals both CO and NCO are happening by double Hollyday’s Junction resolution. However, recently, the study performed by Cole and colleagues [4] strongly indicates, that in case of NCO, there is rather a different mechanism, e.g. Synthesis-dependent strand annealing (SDSA). The gene conversion events together with CO contribute to the evolution of genome. Gene conversion copies a short (tens to hundreds of bases) segment of DNA from
Semantic Interoperability in Biomedicine and Healthcare
14
Gergelits V. – On Gene Conversion Properties
the donor chromatid to the recipient homologous chro(donor) to another (recipient), resulting in the rematid. The original segment of DNA of the recipient placement, insertion, or deletion of a DNA sequence chromatid is replaced by the segment of donor chromatid, in the recipient. while the donor chromatid remains unchanged. There are multiple ways how gene conversion con- Reference: [3] tributes to genome evolution. I will provide three exam- SNOMED CT: not found ples: MeSH: D005785 1. The gene conversion causes GC bias. The onedirectional transfer of polymorphism is more com- ICD10: not found mon for bases G and C than for bases A and T. It is thought that this GC bias may be an adaptation to Homologous Recombination the high rate of methyl-cytosine deamination which can subsequently lead to transitions of C to T. Definition: The process by which segments of DNA are 2. Gene conversion is a major player in the centromere evolution, where the rate of CO is generally much lower than in other parts of genome [7]. 3. The gene conversion plays an important role in so called hotspot paradox. The question is how can recombination hotspots, determined in-trans by Prdm9 gene, exist? Being recombination hotspots, they are more prone to DSB creation, which are repaired by process leading to gene conversion. The DSB repair leads to the modification of binding site, which can no longer be a binding site. Recently, the study by Cole and colleagues [4] provided a partial explanation for this paradox. It was shown that the erosion of binding sites occurs only in 20% of gene conversions during NCO, while only the surroundings of the binding sites are changed in the rest of the cases.
exchanged between two DNA duplexes that share high sequence similarity. Reference: [1] SNOMED CT: not found MeSH: D059765 ICD10: not found
Double-Strand Break Definition: Breaks in opposite DNA strands that lie within 10-20bp of each other. Synonyms: DSB Reference: [1] SNOMED CT: not found
Application in Biomedicine and Healthcare
MeSH: D053903
NCO and gene conversion are processes, which occur frequently during meiosis, an essential process of all sexually reproducing organisms. NCO thus closely relates to patogenesis in human. Gene conversion has been implicated in at least 18 human diseases, e.g. Campomelic dysplasia, Hypergonadotrophic hypogonadism, Hurler-Scheie syndrome [1]. Based on our unpublished data we assume that gene coversion could be related to mechanisms of hybrid sterility.
References
Acknowledgements This paper has been partially supported by the SVV2015-260158 project of Charles University in Prague.
Keywords
ICD10: not found
[1] Chen, J. M., Cooper, D. N., Chuzhanova, N., F´ erec, C., & Patrinos, G. P. (2007). Gene conversion: mechanisms, evolution and human disease. Nature Reviews Genetics, 8(10), 762-775. [2] Williams, A. L., Genovese, G., Dyer, T., Altemose, N., Truax, K., Jun, G., ... & Przeworski, M. (2015). Non-crossover gene conversions show strong GC bias and unexpected clustering in humans. eLife, 4, e04637. [3] Assis, R., & Kondrashov, A. S. (2012). A strong deletion bias in nonallelic gene conversion. PLoS Genet, 8(2), e1002508e1002508. [4] Cole, F., Baudat, F., Grey, C., Keeney, S., de Massy, B., & Jasin, M. (2014). Mouse tetrad analysis provides insights into recombination mechanisms and hotspot evolutionary dynamics. Nature genetics, 46(10), 1072-1080. [5] Jeffreys, A. J., & May, C. A. (2004). Intense and highly localized gene conversion activity in human meiotic crossover hot spots. Nature genetics, 36(2), 151-156.
Gene Conversion
[6] Szostak, J. W., Orr-Weaver, T. L., Rothstein, R. J., & Stahl, F. W. (1983). The double-strand-break repair model for recombination. Cell, 33(1), 25-35.
Definition: Gene conversion is a process whereby a DNA sequence is copied from one segment of the genome
[7] Shi, J., Wolf, S. E., Burke, J. M., Presting, G. G., Ross-Ibarra, J., & Dawe, R. K. (2010). Widespread gene conversion in centromere cores. PLoS Biol, 8(3), e1000327.
Semantic Interoperability in Biomedicine and Healthcare
15
Hruˇskov´a L. – Mutation Detection of Collagen Type I Genes
Mutation Detection of Collagen Type I Genes Lucie Hruˇskov´ a1 1
Department of Pediatrics and Adolescent Medicine, First Faculty of Medicine, Charles University in Prague, Czech Republic
Correspondence to: Lucie Hruˇskov´ a First Faculty of Medicine, Charles University in Prague Address: Kateˇrinsk´ a 32, 121 08 Prague 2 E–mail:
[email protected]
Aims of Research
1(I) chains are not incorporated into the collagen type I. This leads to reduced production of final proteine.
Collagen type I is a major structural protein of connective tissue, especially of bones and skin. It is a heterotrimer composed of two copies of alpha1 (I) and one alpha2 (I) chains encoded by COL1A1 and COL1A2 genes. It generates interactions with other molecules of extracellular matrix (cartilage oligomeric protein, integrins decorin, collagen type V, phosphophoryn etc.). Due this interactions collagen type I increases integrity of connective tissue [1, 2].
OI types II–IV originate in structural changes of COL1A1 and COL1A2 genes. 80% of these forms is caused by glycine substitution (it is the most important amino acid of alpha chains that occurs in 338 Gly-X-Y repetitive motives), 20% are results of splice site mutations. Severity of these mutations increases toward to C-terminus of gene (alpha chains start aligning into the heterotrimer at C-terminus) [3].
Aim of this research is assembling of molecular genetic data of patients with defective collagen type I production and the comparison of obtained data with phenotypes of these individuals. This study is focused on analysis of DNA samples (using methods Polymerase chain reaction (PCR) and Sanger sequencing) of patients diagnosed with osteogenesis imperfecta, type I–IV.
The molecule of collagen type I contains three multi ligand binding regions in the molecular in which are highly concentrated interactions with other molecules. Mutations of the first region (MLBR1) usually lead to milder phenotypes (OI types I and IV) whereas structural changes of MLBR2 and MLBR3 caused severe or lethal forms of disorder (OI type II and III) [4].
State of the Art
Application in Biomedicine and Healthcare
Genetic changes of collagen type I result either in production of defective molecules of the protein or in decreased gene product activity. Mutations of COL1A1 Molecular genetic analyses of collagen type I genes are genes are associated with phenotypes of osteogenesis imcrucial for identification of the largest mutational specperfecta type I–IV, Ehlers-Danlos syndrome (classical trum in Czech patients. Obtained data will be comtype and type VIIA), Caffey disease, and idiopathic ospared with world databases (Online Mendelian Inheriteoporosis [2]. tance in Man, Human Genome Mutation Database, EnOsteogenesis imperfecta (OI), type I–IV, is a herita- sembl, GeneCards, etc.). This way we could be able to ble bone fragility disorder characterised by short stature, compare the disease etiology of Czech patients with other deformed bones, blue or grey colorit of sclera, dentino- populations or ethnics groups. Further, identification of genesis imperfecta. It occurs in 1 in 15000-20000 births. DNA changes and their impact on clinical picture will be The severity of OI differs from mild (OI I) to lethal (OI helpful for initiating timely and appropriate treatment. type II) forms. Presence of clinical signs differs not only This would be also useful for predictive diagnostics in between individual types of the disease, but also in cohort individuals with suspected defects of collagen type I. In of patients of the same form of OI [3]. other case, if analyses do not detect structural changes of The first type of OI is in most cases result of STOP COL1A1 or COL1A2 genes, we can consider possibility codons in COL1A1 gene which causes the premature ter- of location of causal mutations in other genes associated mination of transcription. Not fully transcripted alpha with collagen type I. Semantic Interoperability in Biomedicine and Healthcare
16
Hruˇskov´a L. – Mutation Detection of Collagen Type I Genes
Acknowledgements
Nonsense Codon
This paper has been partially supported by the SVV- Definition: A codon for which no norma tRNA exists; 2015-260158 project of Charles University in Prague. the presence of a nonsense codon causes termination of translation (ending of the polypeptide chain). The three nonsense codons are called amber, ocher Keywords and opal.
Amino Acid
Synonyms: STOP codon
Definition: A peptide; the basic building block of pro- Reference: Griffiths AJF et al. An introduction to geteins (or polypeptides). netic analysis. Sixth edition. New York, USA: W. H. Freeman and Company; 1996. str. 871. Reference: Griffiths AJF et al. An introduction to genetic analysis. Sixth edition. New York, USA: W. SNOMED CT: not found H. Freeman and Company; 1996. str. 859. SNOMED CT: 52518006 MeSH: D000596
MeSH: D018389 ICD10: not found
ICD10: not found
Deoxyribonucleic Acid Definition: One of two types of molecules that encode genetic information. (The other is RNA. In humans DNA is the genetic material; RNA is transcribed from it. In some other organisms, RNA is the genetic material and, in reverse fashion, the DNA is transcribed from it.
Polymerase Chain Reaction Definition: A technique for amplifying a large number of copies of a specific DNA sequence flanked by two oligonucleotide primers. The DNA is alternately heated and cooled in the presence of DNA polymerase ad free nucleotides so that the specified DNA segment is denatured, hybridized with primers, and extended by DNA polymerase.
Synonyms: DNA Reference: http://www.medicinenet.com/script/ main/art.asp?articlekey=3090
Synonyms: PCR
SNOMED CT: 24851008
Reference: Carey J, White B. Medical genetic. Third edition. St. Louis, Missouri: Mosby; 2003. str. 348
MeSH: D004247
SNOMED CT: 702675006
ICD10: not found
Mutation Definition: A permanent structural alteration in DNA. In most cases, DNA changes have either no effect or cause harm, but occasionally a mutation can improve an organism’s chance of surviving, and the beneficial change is passed on to the organism’s descendants. Typically, mutations are more rare than polymorphisms in population samples because natural selection recognizes their lower fitness and removes them from the population.
MeSH: D016133 ICD10: not found
References [1] Barnes MA, Weizhong Ch, Morello R, Cabral WA, Weis M, Eyre DR et al. Deficiency of cartilage-associated protein in recessive lethal osteogenesis imperfecta. N Engl J Med. 2006; 355(26):2757-2764. [2] Fahiminiya S, Majewski J, Mort J, Moffatt P, Glorieux FH, Rauch F. Mutations in WNT1 are a cause of osteogenesis imperfecta. J. Med. Genet. 2013; 50:345-348.
Reference: http://www.ncbi.nlm.nih.gov/books/ NBK21106/pdf/Bookshelf_NBK21106.pdf
[3] Forlino A, Cabral WA, Barnes AV, Marini JC. 2011. New perspectives on osteogenesis imperfecta. Nat Rev Endocrinol. 2011; 7:540–557.
SNOMED CT: 55446002 (ID found for ”Genetic mutation”)
[4] Sweeney, SM, Orgel JP, Fertala A, McAuliffe JD, Turner KR, Di Lullo GA, Chen S, Antipova O, Perumal S, Ala-Kokko L, Forlino A, Cabral WA, Barnes AM, Marini JC, San Antonio JD. Candidate cell and matrix interaction domains on the collagen fibril, the predominant protein of vertebrates. J Biol Chem. 2008;283:21187-21197.
MeSH: D009154 ICD10: not found
Semantic Interoperability in Biomedicine and Healthcare
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
Statistical Methods for Constructing Gestational Age-Related Charts for Fetal Size and Pregnancy Dating using Longitudinal Data Martin Hynek1,2 , Jeffrey D. Long3,4 , David Stejskal2 , Jana Zv´ arov´ a2,5 1 2
Gennet, Centre for Fetal Medicine and Reproductive Genetics, Prague, Czech Republic
Institute of Hygiene and Epidemiology, First Faculty of Medicine, Charles University, Prague, Czech Republic 3
Department of Psychiatry, Carver College of Medicine, University of Iowa, Iowa City, IA, USA
4
Department of Biostatistics, College of Public Health, University of Iowa, Iowa City, IA, USA 5
EuroMISE Centre, Institute of Computer Science AS CR, Prague, Czech Republic
Correspondence to: Martin Hynek Gennet, Centre for Fetal Medicine and Reproductive Genetics Address: Kosteln´ı 9, 170 00 Prague 7 E–mail:
[email protected]
Introduction The assessment of fetal size and the accurate estimation of gestational age (GA) are of crucial importance for proper pregnancy management. Early detection of fetal growth restriction or macrosomia may decrease associated morbidity and mortality [1, 2]. Precise information on GA may prevent unnecessary obstetric interventions at the time of delivery [3]. The information is almost exclusively based on ultrasound measurements of fetal biometric parameters (e.g. crown-rump length, head circumference, femur length, etc.), which almost invariably increase with GA. The means for evaluating these measurements are age-related reference charts (centile charts) allowing interpretation of obtained fetal measurement in comparison with the expected average measurement in the reference population. A variety of statistical methods for constructing reference charts has been suggested. To allow for elaboration, let y be the fetal measurement of interest, and let t be gestational age. The statistical objective is to estimate the centile, cα (y|t), of the conditional distribution of y (given t) for specified values of α (e.g. α = 0.95 for the (100·α)th = 95th centile). Moreover, cα (y|t) is required to be a smooth function of t [4]. There are two main issues in the estimation of cα (y|t): approximating the distribution of y|t, and smoothing estimated centiles on t. Several approaches have been suggested to tackle these issues, including parametric, semiparametric and nonparametric Semantic Interoperability in Biomedicine and Healthcare
techniques. Detailed overviews and comparisons of different approaches can be found in the literature [5, 6, 7, 8]. Interest is usually in the construction of two types of reference charts: charts of fetal size and dating charts. The former is used to assess fetal size, whereas the latter to predict GA. It is incorrect to use the size charts to estimate GA; proper dating charts should be constructed [9]. If the charts of fetal size model the fetal size as a function of GA, the dating charts are produced in similar fashion to size charts, except that they are based on modelling GA as a function of fetal dimension using the conditional distribution, t|y. An appropriate method for constructing reference charts fulfills certain requirements. Altman and Chitty (1993) state that reference centiles should change smoothly with gestation, and they should provide a good fit to the raw data. It is desirable for the statistical model to be as simple as is compatible with these requirements [9]. The WHO Multicentre Growth Reference Study Group (MGRS) [8] agreed on primary criteria for method selection, which include the ability to: • estimate precisely outer centiles, • estimate centiles simultaneously in such a way that they are constrained to be ordered (not to cross), • estimate z-score and centiles using direct formulae, • apply continuous age smoothing, and
17
18
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
• account for both skewness and kurtosis when neces- previous studies that the distribution of fetal dimensions sary. is close to normal for any GA [12]. In addition, the mean is a non-linear function of GA for most fetal dimensions, In addition, the secondary criteria were: and so too the SD. Therefore, the most used method for construction of fetal reference charts is linear regression • the ability to assess fit to the data, using fractional polynomials (FPs)([13]), fitted separately for the mean and the SD, assuming normality at each GA, • ease of explanation and clear documentation, but non-linearity as a function of GA [4, 13, 14, 15]. • useful for application to different anthropometric First, we will describe the mean and SD method usmeasures, so that WHO growth curves would rely ing conventional polynomials (CPs) for the cross-sectional on a single approach. data, then introduce its modification using FPs, and disThe vast majority of literature dealing with fetal reference charts describes statistical methods suitable for cross-sectional data, ie. each fetus contributes only one observation to a reference sample. Because single-visit data is easiest to collect and methodology for computing centiles is straightforward, most published papers in past decades have used cross-sectional datasets to produce reference charts. However, longitudinal data are becoming more common and different techniques must be used for data in which every fetus is measured more than once. Among the challenges, serial measurements of individual fetuses induce correlated data that needs to be accounted for in the statistical modeling. There is also the question of how to compute degrees of freedom (df ), especially when the number of repeated measurements varies among the fetuses [9]. The aim of this article is to compile and propose statistical methods for constructing age-related fetal reference charts using longitudinal data. The method will be used to produce reference charts of fetal size and charts for pregnancy dating for the Czech population in a future analysis. The study will be based on a large dataset collected during routine ultrasound scans in the Fetal Medicine Centre Gennet in Prague. Because a considerable proportion of our fetuses are measured more than once during the course of pregnancy, we are focusing on accommodating currently available methods for construction of fetal reference charts to longitudinal data. This article focuses solely on statistical methods. The equally important issues concerning sample size, sample selection, inclusion and exclusion criteria, and data collection, are not discussed here and can be found in the relevant literature [9, 10, 11, 12].
Statistical methods When searching for a suitable statistical method that fulfills the requirements mentioned above, we have to realize two specifics of fetal biometry. First, it is well established that mean fetal dimensions increase monotonically during pregnancy. The between-subject variability of fetal dimensions also tends to increase (spread out over time), which is summarized by the standard deviation (SD) over age. It is crucially important to consider not only the relation between the mean and GA, but also the relation between the SD and GA [9]. Second, we know from many
cuss methods for checking the goodness-of-fit. Finally, we will propose an extension of the method to longitudinal data.
Cross-sectional data Mean and SD model The original mean and SD approach represents parametric modeling proposed by Altman [14] and Royston and Wright [15]. The method assumes normally distributed fetal measurements at each GA and uses CP regression to model the mean and the SD as a function of GA. A desired centile curve is then constructed with the centile defined as cα = µ + kσ, (1) where k is the corresponding centile of the standard normal distribution and µ and σ are the mean and SD, respectively, at the required GA for the reference population. The original method consists of several steps. First, CP regression is used to describe the mean as a function of GA. The authors suggest starting with a cubic polynomial, and proceed to reduce the number of terms in a top-down manner if the coefficient of the highest term is not significantly different from zero. Once a suitable mean model is selected, a separate analysis is performed to model the variability around the mean. The authors assume that if the variable has a normal distribution at all ages, then the residuals should have a normal distribution and the absolute values of residuals should have the half normal distribution. The p mean of the half standard normal distribution equals 2/π. The scaled absolute residuals are the absolute residuals multip plied by π/2. If we regress the scaled absolute residuals on GA, the predicted values from this model give agespecific estimates of the SD of the residuals, and hence of y. Similarly, CP regression is used to estimate the appropriate relationship in the same manner as for the mean. Finally, because SD increases with GA (heteroskedasticity), weighted least-squares regression can be used, with weights being equal to the reciprocal of the square root of the estimated SDs [16]. However, Altman and Chitty report that the effect of the weighting is almost always rather small and CP regression might suffice [9]. Semantic Interoperability in Biomedicine and Healthcare
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
Fractional polynomials CPs, which were used in the original Altman’s and Chitty’s mean and SD method, suffer from several wellknown limitations. Low order polynomials offer only a few curve shapes, higher order polynomials may fit badly at the extremes, and polynomials do not have asymptotes [13]. Royston and Altman (1994) introduced an extended family of curves known as FPs [13]. FPs are similar to CPs in that their time transformations are power functions, and FPs subsume CPs as a special case. However, the powers in FPs are allowed to be negative numbers and fractions. Possible exponents are usually chosen from a small preselected set of integer and non-integer values: S = {−3, −2, −1, −0.5, 0, 0.5, 1, 2, 3}, where 0 indicates the natural log transformation. The elements in S are the main values of Tukey’s ladder of re-expressions, used for general curve-fitting problems [17], but different sets are possible. Attractive features of FPs include parsimony (they provide similar fit as polynomials but with fewer terms), a wide range of curve shapes, and the ability to approximate asymptotes [4, 18]. To fix ideas, consider the equation for the FP of order m, denoted as FPm, with power terms p = (p1 ≤ . . . ≤ pm ), m X φ∗ (t; p) = β0 + φm (t; p) = βj hj (t), (2) j=0
where h0 (t) = 1 and p t j hj (t) = hj−1 (t) log t
if pj = 6 pj−1 if pj = pj−1
(3)
for j = 1, . . . , m, and t0 ≡ loge (t) [4, 13]. A further constraint is t > 0, to ensure the FP transformation will always be defined. For example, a first-order FP (FP1) with p1 = 0 is β0 + β1 log(t). A second-order FP (FP2) with p1 = −2 and p2 = 1 is β0 +β1 t−2 +β2 t, and a third-order FP (FP3) with p = (0, 2, 2) is β0 + β1 log t + β2 t2 + β3 t2 log t. FPs with m ≤ 2 offer a wide variety of non-linear (and linear) curves. Thus, FP1 or FP2 regression models suffice for many practical applications [4]. FP1 functions are always monotonic, whereas FP2 functions are monotonic or non-monotonic (convex/concave), with the quadratic polynomial being a special case (p1 = 1, p2 = 2). Because the biological nature of fetal growth is largely monotonic, it seems sensible that FP1 models would be sufficient for constructing fetal reference charts. However, FP2 function may also have application, especially if development is considered over a relatively wide epoch. Because FP2 curves cover a wider class of trends, usually both FP1 and FP2 are considered when fetal charts are produced. Indeed, previous experience with FPs in fetal charts indicates that FP2 models were the best-fitting in many cases [19, 20, 21]. When using FPs in regression, how do we select the best model(s) among all the possible candidates? We Semantic Interoperability in Biomedicine and Healthcare
assume throughout that maximum likelihood (ML) estimation is used, with the key sample quantity being −2 times the maximized log-likelihood, known as the deviance. When all the candidate FP models are of the same order (e.g., FP1), then the best model is simply the one with the smallest deviance [4]. However, when dealing with FP models of different order, an alternative strategy is required. The reason is that the deviance will always decrease even when a worthless predictor is added to the model [18]; thus, more complex models would be preferred to less complex ones. To address the issue of inference with models of different order, Royston and Sauerbrei [4] note that under some general assumptions, the deviance difference between a null model with [β1 , . . . , βm ]T = 0 and a FP model of order m is approximately asymptotically distributed as χ2 on df = 2m. Similarly, the deviance difference between FPm and FP(m − 1) models is distributed approximately as χ2 on df = 2, under the null hypothesis that the additional β in the FPm model is zero [4]. Thus, the likelihood ratio test (LRT) can be used to compare models. The originally proposed sequential procedure by Royston and Altman [13] had a series of steps, comparing FP2 with FP1 models, then FP1 with the linear model (p1 = 1), and then finally the linear model with the null model (β1 = 0). However, this approach can greatly increase type I error probability when t is uninfluential and may sometimes give inconsistent results [4]. Therefore, Royston and Sauerbrei [4] recommend an alternative approach, which is a closed test procedure that preserves type I error probability at a chosen level α (typically α = 0.05). The procedure runs as follows: 1. Test the best FP2 model at the level of α against the null model (all βj>0 = 0) using df = 4. If the test is not significant, stop and conclude that t is a worthless predictor. Otherwise continue. 2. Test the best FP2 model against a linear model (p1 = 1) using df = 3. If the test is not significant, stop and the final model is the linear model. Otherwise continue. 3. Test the best FP2 model against the best FP1 model using df = 2. If the test is not significant, the final model is FP1; otherwise the final model is FP2. Modeling the mean and SD with FPs is performed in the same way as described above. The only difference is that FPs are used instead of CPs to model the mean and the SD of y|t. The evaluation of goodness-of-fit As repeatedly emphasized in the methods literature [6, 8, 9, 12], it is absolutely necessary to assess how well the final models fit the data. Many diagnostic tools have been proposed. Here we present a list of the most frequently mentioned methods, including those recommended by the WHO MGRS Group [8]:
19
20
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
• visual inspection of the shape of the centile curves leading to biased standard errors and related inferential plotted together with data points, statistics. Clearly, traditional regression is inappropriate for longitudinal data and an appropriate alternative needs • empirical and fitted centiles plotted on top of each to be used [18]. other, • comparison of observed percentages that occur Linear mixed effect regression above and below estimated centiles against exLinear mixed effects regression (LMER) is a method pected, for the analysis of clustered (or nested) data that includes • scatterplot of the distribution of residuals in z- longitudinal data [27]. Correlation due to repeated meascores, sures is accounted for by subject-level terms known as random effects, which allow for individual variation in change • normal Q-Q plot of residuals in z-scores, curves. LMER also has fixed effects, which are similar to • detrended normal Q-Q plot (worm plot) of residuals the traditional regression coefficients, constant among inin z-scores, i.e., plotting empirical quantiles minus dividuals, and index group-level change [18]. The presence the normal quantiles against the normal quantiles, of fixed effects and random effects gives rise to the term LMER. • normality test of residuals in z-scores or their comIn the longitudinal context, an individual’s random efbinations, e.g., the Q-statistic [22] (a combination fects indicate the deviation of their regression line from of tests for the four moments of the distribution, the group regression line. When straight-line change is the modified D’Agostino tests [23] and Shapiro-Wilk considered, we can have two different (but possibly corretest), lated) random effects: random intercepts (the difference between individuals’ intercepts and the group intercept) • smoothed residuals as a function of t plotted as a and random slopes (the difference in rate of change for solid line together with a pointwise 95% confidence individuals compared to the group slope). In the case of interval. Such plots are useful in identifying regions fetal biometry it seems appropriate to consider both ranof explanatory variables within which the model did dom effects. The biological nature of fetal growth suggests not adequately fit the data. Royston and Sauerbrei that not only are there differences in the size at the start[4] present their results using an univariate runninging point of the observation period (represented by the line smoother implemented in the Stata software by intercept), but also in the rate of growth (represented by Sasieni [24], and similar results can be produced in the slope). the R software using Friedman’s supersmoother [25]. An advantage of LMER is that all subjects need not The above tools may be applied to the final curve, after be measured at the same times nor have the same numbers of observations. Time imbalance is a quality of our smoothing across GA, or locally, at age-specific levels. data set, as the individual fetuses differ in the number of measurements (from one to about ten), and the timing Extension to longitudinal data of measurements differ between the fetuses (some closely Longitudinal data arise from observing the same indi- spaced and others more distantly spaced). In our view, viduals repeatedly over time [26]. As already mentioned, LMER constitutes a desirable approach because of the our planned study will aim to construct fetal reference following: charts for the Czech population using data from our cen• LMER can accommodate missing data. Subjects tre. Some fetuses are examined only once during pregwho have data for at least one time point can be innancy. However, a considerable proportion of fetuses is cluded in the analysis. (ML estimation is unbiased measured more than once (up to ten times). A purely assuming the missing data mechanism is ignorable.) cross-sectional data analysis approach would clearly be incorrect because serial measurements on an individual • LMER is very flexible regarding the structure of the fetus are highly correlated and the effective sample size is data. Timing of observations can vary among sublikely to be nearer to the number of fetuses than to the jects, and the distance between time points need not number of observations. The typical assumption of tradito be equal. tional regression about the independence of observations is certainly violated. Assuming random sampling of fetuses, • LMER provides control over the number and nature it is plausible that subjects are independent, but the obof terms used to model change over time. CPs, FPs, servations that are nested within subjects are dependent. and a wide variety of other transformation (e.g., In cross-sectional data analysis the focus is on betweensplines) can be accommodated. subjects variability and differences, but longitudinal data analysis adds the dimension of within-subjects variability. Turning to the specifics of LMER, suppose yij is the Applying traditional regression to longitudinal data is in- response score for the ith individual (i = 1, . . . , N ) at correct because the within-subject variability is ignored the jth time point (j = 1, . . . , ni ), m is the number of Semantic Interoperability in Biomedicine and Healthcare
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
predictors of the fixed effects, and l is the number of pre- that the correlation (covariance) of the random effects imdictors of the random effects (both excluding intercept). plies that the variability of fetal measurements can inThe general form of the classical LMER model is [28, 29] crease over time, which is a characteristic that is important to model. In the case of FP2 or even higher order yi = Xi β + Zi bi + εi , (4) models, there is a decision that needs to be made in regard to including additional random effects. The inclusion where yi is an ni × 1 individual time series vector, Xi is a of many random effects can sometimes lead to estimation known ni × (m + 1) design matrix for the fixed effects, β problems and lowered efficiency, especially if the sample is a (m + 1) × 1 vector of unknown fixed effects, Zi is an size and/or the number of observations are limited. When ni × (l + 1) known design matrix for the random effects, selecting among models with a different number of fixed bi is a (l + 1) × 1 vector of unobserved random effects, and effects, it is important that the comparison models have εi is an ni × 1 vector of unobserved random errors. More the same number of random effects in order to make the explicitly, if the predictors are FPs of time (t), source of misfit unambiguous. Therefore, comparison of FP1 and FP2 models should involve both having only two εi1 yi1 random effects. In anticipation of this type of comparison, .. .. . . we consider models with only two random effects. , εi = εij , In general, FP1 for LMER is defined as y (5) yi = ij . . .. .. (10) yij = (β0 + b0i ) + (β1 + b1i )tpij1 + εij . εini yini FP2 has two power terms, p1 and p2 , and with p1 ≤ p2 . β0 When there are distinct powers, p1 < p2 , and the FP2 1 tpi11 . . . tpi1m β1 .. . . . .. .. .. , β = . , (6) with two random effects is Xi = . .. p1 pm 1 tini . . . tini yij = (β0 + b0i ) + (β1 + b1i )tpij1 + β2 tpij2 + εij , (11) βm In the case of equal powers, p1 = p2 , the model is b0i 1 tpi11 . . . tpi1l b1i .. .. , b = .. Zi = ... (7) yij = (β0 + b0i ) + (β1 + b1i )tpij1 + β2 tpij1 log tij + εij . (12) .. . . . . i . 1 tpin1i . . . tpinl i bli In the above FP2 models, the random slopes are based The columns of Zi are typically a subset of the columns on the first type of FP transformation (ie., on p1 ), but of Xi , so l ≤ m. For FP1, it is usual that l = m = 1 they can just as easily be based on the second (ie., on p2 ). and random intercepts and random slopes are specified. Furthermore, it should be noted that p1 may be a differThe typical normality and related assumptions are sum- ent value among the FP1 and FP2 models when many comparisons are made as discussed below. marized as In a pure exploratory analysis, we want to examine all bi ∼ N (0, G), εi ∼ N (0, Ri ), with bi ⊥εi , (8) FP1 and FP2 models. In this scenario, all the values in S (9+1)! = 45 are considered yielding nine FP1 models and 2!(9−1)! ie., the residual errors εi for the same individual are in- FP2 models. These 54 models are fitted using full ML dependent of the random effects bi . We further assume rather than REML because they differ only in the numRi = σε2 Ini , where Ini is an ni ×ni identity matrix, and G ber of fixed effects [18, 28]. is the (l + 1) × (l + 1) unknown variance-covariance matrix for the random effects and the variances on the diagonal Multimodel inference using Akaike’s information are non-negative. The fixed effects and variance compo- criterion nents are typically estimated with full ML or restricted ML (REML) [18, 28]. Given a total of 54 FP1 and FP2 fitted models, the For example, the simple LMER linear trend model selection of the best (or a best subgroup) is a challenging with random intercept and random slope uses an FP1 with task. As already mentioned, in the case of cross-sectional p1 = 1 (l = m = 1), data and traditional regression, Royston and Sauerbrei [4] proposed the use of the LRT. The LRT is problematic yij = (β0 + b0i ) + (β1 + b1i )tij + εij . (9) to use in this context because many comparisons will be made among models with the same numbers of paramFractional polynomials eters, and the LRT can only be used for nested model testing. Therefore, for the comparison of our 54 LMER As shown above, FPs can be directly incorporated into models, we suggest an alternative strategy known as mulLMER. Because of the observed variation in fetal growth, timodel inference [30]. Multimodel inference is used rouit seems appropriate to include both random intercepts tinely for model selection in ecology and wildlife biology and random slopes in our LMER models. It can be shown [31, 32], and has been recently introduced for longitudinal Semantic Interoperability in Biomedicine and Healthcare
21
22
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
data analysis in the behavioral sciences [18] and medicine [33, 34]. Multimodel inference is a statistical strategy based on a new class of information-theoretic (I-T) approaches. The I-T methods allow one to select the best model from an a priori set, rank and scale the models, and include model selection uncertainty into estimates of precision [35]. I-T refers to the methods having a basis in KullbackLeibler (K-L) information theory. K-L information represents the information loss when a particular model is used to approximate the model of full reality. The model of full reality is considered to be unknowable, so the goal here is to select the model in the set of possible models that minimizes K-L information loss (ie., the model that is closest to full reality). The methods considered here are due to Akaike [36] who found a formal relationship between KL information and likelihood theory. He found that the maximized log-likelihood value is a biased estimate of K-L information and that this bias is approximately equal to the number of estimable parameters in the approximating model. The expected, relative K-L information can be estimated by Akaike’s information criterion (AIC) defined as ˆ AIC = deviance + 2K = −2 log(L(θ|y)) + 2K,
(13)
ˆ where log(L(θ|y) is the value of the maximized loglikelihood over the unknown parameters (θ), given the data y, and K is the number of estimable parameters. The AIC penalizes the deviance by 2 times the number of estimated parameters which helps to guard against improving fit simply by adding a worthless predictor. In the case of small samples, a bias-adjusted form of the AIC should be used, the AIC-corrected (AICc)([37]). For longitudinal data, the AICc is computed as 2 · K · (K + 1) , AICc = AIC + P N i ni − K − 1
models. Let ∆h denote the difference for the hth model. Then ∆h = AICch − AICcmin , (15) where h = 1, . . . , H, with H being the number of models, and AICcmin is the smallest AICc value in the set of H models. The best fitting model has ∆ = 0 and the worst fitting model has the maximum ∆. For any set of FP models, there is likely to be a subgroup that has similar good fit and other models with relatively poor fit. There are no definite cutoffs for distinguishing plausible and implausible subgroups of models. However, rough guidelines are the following: models with ∆ ≤ 2 have substantial plausibility support, those with 4 ≤ ∆ ≤ 7 have considerably less support, and models with ∆ > 10 have essentially no support [31]. Another tool for model comparison is a scaled version of ∆, known as the weight of evidence (or the Akaike weight) [31]. The weight of evidence for the hth model, Wh , is a probability scaling of ∆h , exp(−0.5 · ∆h ) . Wh = PH h exp(−0.5 · ∆h )
(16)
Given the data, the set of models, and the unknowable true model, Wh indicates the probability that the hth model is the best approximating model [18]. The model with the largest W is the best approximating model, though many models can have close large values. Similar to ∆, there are no rigid W cutoffs for model plausibility. However, W = 0.90 and W = 0.95 have been suggested as benchmarks [31]. In the event no single model has a particularly large probability, it is convenient to form a confidence set of the largest-valued models whose probabilities sum to 0.90 or 0.95.
A third tool for model comparison is the evidence ratio [31], which expresses how much more likely the best model (14) is than a given model. Suppose that Eh is the evidence ratio for the hth model. Then
where the summation is the total number of observed time points. As the sample size increases, the AICc approaches the AIC. Because of this, the AICc is recommended for both small- and large-sample situations [18]. The AICc can be used to rank-order a set of models in terms of their relative fit and plausibility. If one computes AICc for each of the H models from the a priori selected set, the model with the smallest AICc value is the best approximating model, ie., the closest to the unknowable truth. The AICc is only interpretable as a relative fit index. It is relative because we do not know the true model, and we cannot estimate its distance from a candidate model. However, by comparing different candidate models, we can measure how much better one approximating model is compared to another, ie., we can measure effect size. Several tools have been proposed to aid in model comparison based on the AIC (and AICc). One tool for model comparison is the difference between each candidate model and the best fitting of all the
Eh =
Wmax , Wh
(17)
where Wmax is the maximum weight of evidence in the set. Since Eh is a ratio of probabilities, it also can be viewed as the odds that model h is not the best approximating model. The best fitting model has E = 1, and all other models have larger values. Having introduced three AICc tools for model comparison, there are some cautions to mention. Reiterating a previous point, the models to be compared should have the same number of random effects (and associated variance components), otherwise the AICc can be a biased guide in model selection [38, 39]. A second point is that, despite explicitly taking into account the number of predictors, AICc tends to favour complex models [40]. The AICc (and AIC) is not consistent in the sense that increasing the sample size will increase the probability of identifying the true model. This is not a problem if the researcher beSemantic Interoperability in Biomedicine and Healthcare
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
lieves that a true model is infinitely complex and that the AICc tools will help to identify the best approximation of this complexity (from among the candidate models). However, if a researcher believes there is a true model to be discovered, the tools can lead to over-fitting, which is the selection of a model that has too many parameters. Thus, the researcher’s philosophical outlook can influence the degree to which the I-T methods might be useful. Alternative indexes can be used when the goal is to select the true model (see [41]). Multimodel inference is a natural way to deal with the multiple working hypotheses that often arise in applied medical research, especially with exploratory observational data. After a set of candidate models is a priori selected and fitted, the AICc and the related tools can be used to rank order the models in terms of plausibility and used to assess the relative effect size. An important point to bear in mind is that the results deeply depend on which models are selected into the set of candidate models; if none have merit, the models are still ranked. Thus, it is always important to bring substantive information to bear in assessing the worth of the model(s) estimated to be the best. Standard statistical methods can also be used to cope with this matter, for example, indexes of absolute effect size (e.g. R2 ) and the analysis of regression residuals. Multimodel inference stands in contrast to the more popular method of null hypothesis statistical testing (NHST). Because multimodel inference may be new to many readers, it is perhaps worth pointing out some of the major distinctions of the AICc tools relative to NHST. • In the I-T paradigm, the long-run frequentist principles apply to model estimation rather than decisions about hypotheses, as in NHST. Therefore, adjustment for multiple comparisons is irrelevant in multimodel inference. • The AICc tools can be used to compare multiple model simultaneously, whereas NHST only considers two models at a time.
• Tools such as the evidence ratio, allow one to determine evidence for a model, whereas NHST does not.
Construction of fetal reference charts Having reviewed several methods, let us summarize the overall approach for constructing fetal reference charts from longitudinal data. The mean and SD model technique is applied when the mean and the SD of fetal biometric measurement y is modeled as a function of GA and FP functions are used to establish the relationship of y on GA. The set of 54 FP1 and FP2 LMER models is fitted to the data. Then multimodel inference using one of the I-T tools like the weight of evidence is used to select the best approximating model from the set. After the final model is selected, its absolute goodness-of-fit is thoroughly assessed. Finally, reference centile charts are constructed from the estimated best model.
Conclusion The construction of fetal reference charts requires an appropriate statistical methodology, otherwise estimated centiles may be incorrect and may lead to false clinical conclusions regarding fetal development. Previous experience from many studies tells us that the distribution of fetal size is close to normal for any GA. Therefore, the most frequent method for the construction of fetal reference charts is the parametric approach with FP regression functions for the mean and SD of each fetal measurement. This article suggests how this method can be extended to longitudinal data using FPs in LMER. The presented approach includes ML estimation for fitting firstand second-order FP models, and multimodel inference using AICc and related tools as a suitable strategy for model selection.
Acknowledgements
The work was supported by the grant SVV-2015• The AICc tools can be used for comparison of nested 260158 of Charles University in Prague. and non-nested models. • The AICc tools do not rely on arbitrary cutoffs for Keywords decisions about models, such as the p ≤ 0.05 cutoff in NHST. Models can simply be rank ordered and Deviance inferences made based on the ranks without necessarily relegating any models to the dustbin of in- Definition: Deviance is a quality of fit statistic for a significance. model estimated with maximum likelihood methods. It represents a generalization of the idea of using the • The AICc tools provide a clear measure of effect size, sum of squares of residuals in ordinary least squares whereas NHST does not. to cases where model-fitting is achieved by maximum likelihood. The deviance for a model based on • The AICc tools do not require the assumption that a dataset y, is defined as: at least one of the candidate models is true, whereas ˆ NHST does (the null model is assumed to be true). −2 log(L(θ|y)) Semantic Interoperability in Biomedicine and Healthcare
23
24
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
ˆ is the value of the maximized logwhere log(L(θ|y) likelihood over the unknown parameters (θ), given the data y. Reference: Nelder JA, Wedderburn RWM. Generalized linear model. J R Statist Soc A 1972; 135: 370–384. SNOMED CT: no matching results found MeSH: no matching results found ICD10: no matching results found
Kullback-Leibler information
complexity of the model. ˆ AIC = deviance + 2K = −2 log(L(θ|y)) + 2K, where K is the number of estimated parameters in the model and L is the maximized value of the likelihood function for the model. Given a set of candidate models for the data, the preferred model is the one with the minimum AIC value. Hence AIC rewards goodness of fit (as assessed by the likelihood function), but it also includes a penalty that is an increasing function of the number of estimated parameters. The penalty discourages overfitting, as increasing the number of parameters in the model always improves the deviance. The AIC offers a method of rank-ordering models for comparison and indexes the models’ relative effect sizes in terms of distance from the truth.
Definition: Described by S. Kullback and R. A. Leibler in 1951. Let f denote the full reality or truth, and let g denote an approximating model, a probability distribution. Kullback-Leibler information is a measure of the distance between f and g, and is defined Synonyms: AIC for continuous functions as the integral Reference: 1. Long J. Longitudinal data analysis for the Z f (x) behavioral sciences using R. Thousand Oaks, Calif.: dx I(f, g) = f (x) log Sage, 2012. 2. Akaike H. Information theory as g(x|θ) an extension of the maximum likelihood principle. where f and g are n-dimensional probability distriIn Petrov BN, Csaki F (Eds). Second international butions. K-L information, denoted I(f, g), is the insymposium on information theory. Budapest, Hunformation loss when model g is used to approximate gary: Akademiai Kiado, 1973. 3. Burnham KP, Anf . Then, the best model loses the least information derson DR, Huyvaert KP. AIC model selection and relative to other models in the set; this is equivamultimodel inference in behavioral ecology: some lent to minimizing I(f, g), over the set of models of background, observations, and comparisons. Behav interest. Ecol Sociobiol 2011; 65: 23–35. Synonyms: K-L information, Kullback-Leibler diver- SNOMED CT: no matching results found gence, information divergence, information gain, relMeSH: no matching results found ative entropy Reference: Burnham KP, Anderson DP. Multimodel in- ICD10: no matching results found ference: Understanding AIC and BIC in model selection. Sociol Methods Res 2004; 33(2): 261–304. Degree of freedom SNOMED CT: no matching results found MeSH: no matching results found ICD10: no matching results found
Akaike information criterion
Definition: In the context of likelihood ratio testing, the number of degrees of freedom is a parameter that identifies the specific reference chi-squared distribution used for hypothesis testing. It is the number of parameters that differ when comparing a simpler reduced model to a more complex full model. It is also a parameter used in other probability distributions related to the chi-square distribution, such as the Student distribution and the Fisher distribution. In another context, the number of degrees of freedom refers to the number of linearly independent terms involved when calculating the sum of squares based on n independent observations. The term degree of freedom was introduced by R. A. Fisher in 1925.
Definition: Akaike information criterion is a measure of the relative quality of a statistical model for a given set of data providing a means for model selection. AIC is founded on Kullback-Leibler information theory, and it offers a relative estimate of the information lost when a given model is used to represent the process that generates the data. In addition, the AIC is an unbiased estimate of predictive accu- Synonyms: df racy, which is the ability of a model to predict new data. In doing so, it attempts to balance the trade- Reference: 1. Fisher RA. Applications of ”‘Student’s”’ off between the goodness of fit of the model and the distribution. Metron 1925; 5, 90–104. 2. Dodge Y. Semantic Interoperability in Biomedicine and Healthcare
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
The Concise Encyclopedia of Statistics. Germany: Springer-Verlag, 2008. SNOMED CT: no matching results found MeSH: no matching results found ICD10: no matching results found
Inference Definition: Inference is a form of reasoning by induction performed on the basis of information collected on a sample. It is the logic of generalizing sample information to the relevant associated population. Reference: Dodge Y. The Concise Encyclopedia of Statistics. Germany: Springer-Verlag, 2008. SNOMED CT: no matching results found MeSH: no matching results found ICD10: no matching results found
References [1] Boulet SL, Salihu HM, Alexander GR. Mode of delivery and birth outcome of macrosomic infants. J Obstet Gynaecol 2004; 24: 622–629. [2] Fang S. Management of preterm infants with intrauterine growth restriction. Early Hum Dev 2005; 81: 889–900. [3] Hall MH, Carr-Hill RA. The significance of uncertain gestation for obstetric outcome. Br J Obstet Gynaecol 1985; 92: 452–460. [4] Royston P, Sauerbrei W. Multivariable model-building: A pragmatic approach to regression analysis based on fractional polynomials for modelling continuous variables. Chichester, England: John Wiley, 2008. [5] Wright EM, Royston P. A comparison of statistical methods for age-related reference intervals. J R Statist Soc A 1997; 160: 47–69. [6] Silverwood RJ, Cole TJ. Statistical methods for constructing gestational age-related reference intervals and centile charts for fetal size. Ultrasound Obstet Gynecol 2007; 29: 6–13. [7] Hynek M. Approaches for constructing age-related reference intervals and centile charts for fetal size. Eur J Biomed Informatics 2010; 6: 43–52.
[13] Royston P, Altman DG. Regression using fractional polynomials of continuous covariates: Parsimonious parametric modelling. Appl Statist 1994; 43: 429–467. [14] Altman DG. Construction of age-related reference centiles using absolute residuals. Statistics in Medicine 1993; 12: 917–924. [15] Royston P, Wright EM. How to construct ’normal ranges’ for fetal variables. Ultrasound Obstet Gynecol 1998; 11: 30–38. [16] Aitkin MA. Modelling variance heterogenity in normal regression using GLIM. Applied Statistics 1987; 36: 332–339. [17] Mosteller F, Tukey J. Data analysis and regression: A second course. New York: Addison-Wesley, 1977. [18] Long JD. Longitudinal data analysis for the behavioral sciences using R. Thousand Oaks, Calif.: Sage, 2012. [19] Chitty LS, Altman DG. Charts of fetal size: limb bones. BJOG 2002; 109: 919–929. [20] Papageorghiou AT, Kennedy SH, Salomon LJ, et al.; International Fetal and Newborn Growth Consortium for the 21st Century (INTERGROWTH-21st). International standards for early fetal size and pregnancy dating based on ultrasound measurement of crown-rump length in the first trimester of pregnancy. Ultrasound Obstet Gynecol 2014; 44: 641–648. [21] Papageorghiou AT, Ohuma EO, Altman DG, et al.; International Fetal and Newborn Growth Consortium for the 21st Century (INTERGROWTH-21st). International standards for fetal growth based on serial ultrasound measurements: the Fetal Growth Longitudinal Study of the INTERGROWTH-21st Project. Lancet 2014; 384(9946): 869–879. [22] Royston P, Wright EM. Goodness-of-fit statistics for agespecific reference intervals. Stat Med 2000; 19: 2943–2962. [23] D’Agostino RB, Belanger A, D’Agostino Jr RB. A Suggestion for Using Powerful and Informative Tests of Normality. Am Stat 1990; 44: 316–321. [24] Sasieni P, Royston P, Cox NJ. Symmetric nearest neighbour linear smoothers. Stata J 2005; 5: 285. [25] Friedman JH, Silverman BW. Flexible Parsimonious Smoothing and Additive Modeling. Technometrics 1989; 31(1): 3–21. [26] Long J, Ryoo Jihoon. Using fractional polynomials to model non-linear trends in longitudinal data. Br J Math Stat Psychol 2010; 63: 177–203. [27] Fitzmaurice GM, Laird NM, Ware JH. Applied longitudinal analysis. New York: Wiley, 2004. [28] Galecki A, Burzykowski T. Linear mixed-effects models using R a step-by-step approach. New York: Springer, 2013. [29] Laird N, Ware J. Random-effects models for longitudinal data. Biometrics 1982; 38: 963–974. [30] Burnham KP, Anderson DR. Model Selection and Multimodel Inference. New York: Springer, 2002.
[8] Borghi E, Onis M, Garza C, et. al. Construction of the World Health Organization child growth standards: selection of methods for attained growth curves. Statist Med 2006; 25: 247–265.
[31] Burnham KP, Anderson DR. Multimodel inference: understanding AIC and BIC in model selection. Socio Meth Res 2004; 33(2): 261–304.
[9] Altman DG, Chitty LS. Design and analysis of studies to derive charts of fetal size. Ultrasound Obstet Gynecol 1993; 3: 378–384.
[32] Burnham KP, Anderson DR, Huyvaert KP. AIC model selection and multimodel inference in behavioral ecology: some background, observations, and comparisons. Behav Ecol Sociobiol 2011; 65: 23–35.
[10] Villar J, Altman DG, Purwar M, et al.; International Fetal and Newborn Growth Consortium for the 21st Century. The objectives, design and implementation of the INTERGROWTH-21st Project. BJOG 2013; 120 (Suppl 2): 9–26. [11] Ioannou C, Talbot K, Ohuma E, et al. Systematic review of methodology used in ultrasound studies aimed at creating charts of fetal size. BJOG 2012; 119(12): 1425–39. [12] Altman DG, Ohuma EO; International Fetal and Newborn Growth Consortium for the 21st Century. Statistical considerations for the development of prescriptive fetal and newborn growth standards in the INTERGROWTH-21st Project. BJOG 2013; 120 (Suppl 2): 71–76.
Semantic Interoperability in Biomedicine and Healthcare
[33] Paulsen JS, Smith MM, Long JD; PREDICT HD investigators and Coordinators of the Huntington Study Group. Cognitive decline in prodromal Huntington: Disease: implications for clinical trials. J Neurol Neurosurg Psychiatry 2013; 84: 1233– 1239. [34] Leoni V, Long JD, Mills JA, et al.; PREDICT-HD study group. 24S-hydroxycholesterol correlation with markers of Huntington disease progression. Neurobiol Dis 2013; 55: 37–43. [35] Burnham KP, Anderson DR. Kullback-Leibler information as a basis for strong inference in ecological studies. Wildl Res 2001; 28: 111–119.
25
26
Hynek M. et al. – Statistical Methods for Constructing Gestational Age-Related Charts . . .
[36] Akaike H. Information theory as an extension of the maximum likelihood principle. In: Petrov BN, Csaki F, editors. Proceedings of the 2nd International Symposium on Information Theory; 1973; Budapest, Hungary: Akademiai Kiado. p. 267–281. [37] Hurvich CM, Tsai CL. Regression and time series model selection in small samples. Biometrika 1989; 76: 297–307. [38] Liang H, Wu H, Zou G. A note on conditional AIC for linear
mixed-effects models. Biometrica 2008; 95: 773–778. [39] Vaida F, Blanchard S. Conditional Akaike information for mixed-effects model. Biometrica 2005; 92: 351–370. [40] Link WA, Barker RJ. Model weights and the foundations of multimodel inference. Ecology 2006; 87: 2626–2635. [41] Kadane JB, Lazar NA. Methods and criteria for model selection. J Amer Statist Assoc 2004; 99: 279–290.
Semantic Interoperability in Biomedicine and Healthcare
ˇarek M. – Significance of Cerebral Folate Deficiency for Development and Progression of Autism Krsiˇcka D., S´
Significance of Cerebral Folate Deficiency for Development and Progression of Autism ˇ arek2 Daniel Krsiˇ cka1 , Milan S´ 1
1st Faculty of Medicine, Charles University, Prague, Czech Republic 2
EuroMISE Mentor Association, Prague, Czech Republic
Correspondence to: Daniel Krsiˇ cka 1st Faculty of Medicine, Charles University Address: Katerinska 32, 121 08, Praha, Czech Republic E–mail:
[email protected]
Aims of Research We aim at better understanding of underlying relations between autism or Autism Spectrum Disorders (ASD) respectively and Cerebral Folate Deficiency Syndrome (CFD), its etiology, pathophysiology and synergistic effects of a depletion of folates in the various critical developmental periods and other subclinical environmental insults or genome-wide predispositions. Some genes associated with ASD or other neurodevelopmental disorders have a metabolic role as they are translated into enzymes. We suppose that lack of folates can negatively influence some metabolic reactions critical for ASD development or progression in similar way like genetic pathologies even if appropriated genes are intact or synergistically strengthen an effect of less important genetic pathologies. Thus dysregulation of the substrates, metabolites or cofactors could result into the similar disorder like a dysfunctional polymorphism or deletion of appropriate gene under normal concentration of compounds. It could be also interesting to investigate further pathologic effects of folate depletion like changes in gene expression due to folate dysregulation, changes in developmental signaling pathways or change in DNA methylation patterns or relations to partial dysfunction of mitochondrions. Our hypothesis is based on a published clinical experience with diagnose and treatment of concomitant CFD by ASD with reported alleviation or even suppression of core ASD symptoms in some cases. Our first goal is to deterministically find relations among folate metabolism and metabolic reactions associated with autism in as widest informational scope as possible and to determine among the many published biochemical, physiological and clinical observations the significant ones for ASD development and progression. This knowledge, when created, could be further investigated in deeper detail for example relevance of the folate depletion for exact prenatal or postnatal developSemantic Interoperability in Biomedicine and Healthcare
mental periods or dependency of the folate depletion to the higher male prevalence of ASD and many others. We can also test how is the gene expression changes under the condition of folate depletion i.e. which genes would be upregulated and whether some of them have, based on present knowledge, some association with ASD. The set of possible experiments is much wider and it is enriched also by an overlap with other diseases like schizophrenia, amnesia or trisomy 21 which seems to be also partially folate-dependent.
State of the Art Etiology of autism or Autism Spectrum Disorders (ASD) is mostly unknown, genetic syndromes only account for an estimated 15% of autism cases. The ASD etiology is mostly multifactorial and due to this, an effective research should also investigate multiple factors. In our recent review [1], we’ve found a total of 351 published cases of CFD with variable ASD reported in 44% of patients and 56% of these cases have been caused by Folate Receptor Auto Antibodies (FRAA), disturbing or even blocking transport of folates across the bloodbrain barrier. We’re especially focused on these ASD cases due to it is quite hard to diagnose a CFD in ASD patients. The only available and completely reliable method is a lumbal puncture with examination of 5-MTHF (5methyltetrahydrofolate) level in cerebrospinal fluid (CSF), not routinely indicated in ASD for its invasiveness. Less invasive methods, such as MR spectroscopy, have not the needed resolution in nmol/L at present. The FRAA assay is available only in a few laboratories in the world. It is therefore possible that a mild CFD in ASD escapes attention in diagnosis and treatment for a long time, and that it may contribute to development and progression of the ASD. Moreover, 29% of found ASD cases with concomitant CFD show that treatment of CFD can alleviate or occasionally even suppress the core ASD symptoms
27
28
ˇarek M. – Significance of Cerebral Folate Deficiency for Development and Progression of Autism Krsiˇcka D., S´
[2, 3, 4, 5, 6, 7]. It couldn’t be omitted that also other experimental studies of ASD treatment based on administration of substances, whose synthesis or concentration is directly or indirectly folate-dependent or folate-controlled or they act as cofactors in folate metabolic pathways have been published [8, 9, 10, 11, 12, 13]. Significantly elevated FRAA results in a typical degenerative, progressive CFD, however significance of slightly elevated FRAA levels or slightly decreased folate levels in CNS only is not yet precisely known for the pathophysiology of ASD. Abnormalities in folate levels can probably contribute to the many of epigenetic disturbances well-connected to etiology and pathophysiology of ASD and other neurodevelopmental and neurodegenerative disorders like abnormal genome methylation, oxidative stress, mitochondrial and neuronal damage or an abnormal immune response. The long term folate depletion itself is probably not needed but acting together with other factors like insignificant genetic mutations, subclinical environmental toxic burden, long-term stress or specific autoantibodies like the FRAA, it could explain and could cause abnormal CNS growth, neuronal differentiation, migration and pruning or activate apoptosis, leading to neurodegeneration. With relation to neural tissue, long-term insult influence is not needed. The influence of short- or medium-term folate depletion in various developmental intervals has not been systematically studied. The prevalence of mild CFD in ASD is not sufficiently mapped, as well as the contribution of individual pathological causes. This is a very small population tested (total 351 published cases). Even the therapeutic effect on ASD symptoms in treatment of mild CFD has not yet been sufficiently documented. Published results clearly show the need of the further research in this area, primarily because the ASD is currently considered as an incurable lifelong disorder of unknown etiology with a high socio-economic burden. In ASD patients with proven CFD have been repeatedly described improvements of neurological symptoms and also in core ASD symptoms. Some publications show the negative correlation between the age of the patient and treatment outcome. Therefore, an early intervention seems to be essential. Present research of ASD underlying causes consists of 2 major streams. The first stream investigates the polygenic influence of many genes and concentrates on the genome itself. The second stream focuses on epigenetic changes and environmental factors. Both research streams comprise a Big Data Challenge like high-volume data processing from personalized whole genome sequencing, huge data set comparison, automation of new information derivation and others. According to several recent research results it seems that a folate-dependent phenotype could exist for the ASD and the folate depletion can also contribute to development or deterioration of other diseases like schizophrenia [14, 15, 16], sudden adult amnesia [17] or trisomy 21 [18]. These phenotypes can consist of an influence of several harmful factors synergistically acting with folate deple-
tion. The genetic component still seems to be important but cannot convincingly explain all the cases and research results. It seems that the environmental factors, forming the epigenetic influences, should be taken into account. These factors can manifest in other way and couldn’t be primarily connected to the ASD in clinical praxis. A folate-dependent ASD phenotype, when found, could help to significantly affect the further clinical research of ASD treatment according to the rules of evidence based medicine. For the effective resolution of phenotypes, screening, early diagnostics and biological therapies for ASD it is necessary to better understand the synergistic effects of several genetic and epigenetic causes, which together manifest as neurobehavioral syndrome, but each independently represents only a subclinical, not manifest problem. Heterogeneous causes which can occur simultaneously and synergistically reinforcing each other if they are not examined all together, make practically impossible to correctly select the test group for a clinical trial. Mixed test groups composed of more mutually different ASD phenotypes would affect the outcome of each trial in a random manner and the results of these studies would have always the significant differences. Gradually, it would be possible to identify only the dominant causes, as appears to be in serious CFD based on FRAA. Conversely, if it were possible to identify and classify particular, separately insignificant, but collectively manifest causes, it would contribute significantly to the identification of novel ASD etiologies. Therefore, identification and classification of each new ASD phenotype is crucial for further progress.
Application in Biomedicine and Healthcare We’ve some preliminary results confirming our hypotheses. There are 2 main practical usages for our research results of folate-dependency of ASD. • prevention • treatment By the ASD as the lifelong, generally incurable disease is crucial to identify the fragile pregnancy or even parents as soon as possible. If there’s a relation between folate depletion and some relatively novel insults like FRAA influencing a developing fetus or child, it must be identified in order to prevent or early compensate the problem before ASD can develop. We suppose that better and deeper understanding of relations among folates and pathophysiology of ASD can expose other genetic, epigenetic or environmental risk factors, which can contribute to significance of the mild folate depletion thus the factors defining a “folate-fragile” ASD phenotype. This information could be used for development of novel population screening, early intervention and overall could lead to reduction of ASD prevalence and its socioeconomic burden. Last but not least utilization of Semantic Interoperability in Biomedicine and Healthcare
ˇarek M. – Significance of Cerebral Folate Deficiency for Development and Progression of Autism Krsiˇcka D., S´
the research results could be the still open possibility to SNOMED CT: 63718003 use folates as a part of biologically-based ASD therapy MeSH: D03.438.733.631.400 also for already developed cases. ICD10: not found
Acknowledgements Folate Receptor 1 This paper has been partially supported by the SVVDefinition: A subtype of GPI-anchored folate receptors 2015-260158 project of Charles University in Prague. that is expressed in tissues of epithelial origin, clinically significant in the plexus choroid due to 3-times Keywords higher cerebrospinal folate concentrations in comparison with blood level.
Autism
Synonyms: FOLR1 receptor Definition: Pervasive neurodevelopmental disorder characterized by impaired communication skills, dys- Reference: https://en.wikipedia.org/wiki/Folate_ receptor_1 functional social interaction, imagination insufficiency and stereotypic behavior. SNOMED CT: not found Synonyms: Autism spectrum autism (inaccurate)
disorder,
childhood MeSH: D12.776.157.530.450.074.500.299.500.500
Reference: https://en.wikipedia.org/wiki/Autism SNOMED CT: 408856003
ICD10: not found
Folate Receptor Autoantibody
MeSH: F03.550.325.125
Definition: IgM or IgG autoantibody against FOLR1 protein disrupting or blocking the high-affinition, ICD10: no exact reference found (only inaccurate F84.0) low-concentration folate receptor FOLR1 significantly expressed on the blood-brain barrier.
Cerebral Folate Deficiency Syndrome
Synonyms: Folate Receptor Antibody, FOLR1 Antibody Definition: Progressive neurodegenerative syndrome, mostly in early childhood, characterized by decreased level of folate compounds in the central ner- Reference: http://www.ncbi.nlm.nih.gov/pubmed/ 23314538 vous system with simultaneously normal systemic levels. SNOMED CT: not found Synonyms: Neurodegeneration due to lack of cerebral MeSH: not found folates ICD10: not found Reference: http://www.ncbi.nlm.nih.gov/pubmed/ 15581159
References
SNOMED CT: not found MeSH: not found ICD10: no exact reference found (only inaccurate E53.9)
Folate Definition: One of biologically active forms of the folic acid mutually different by variable oxidation/reduction status and variable number of glutamic residues. Synonyms: Folic Acid, Pteroylglutamic Acid, Vitamin B9 Reference: https://en.wikipedia.org/wiki/Folic_ acid Semantic Interoperability in Biomedicine and Healthcare
[1] D. Krsiˇ cka, M. Vlˇ ckov´ a, and M. Havlovicov´ a, “The Significance of Cerebral Folate Deficiency for the Development and Treatment of Autism Spectrum Disorders,” Int. J. Biomed. Healthc., vol. 1, no. 1, 2015. [2] V. T. Ramaekers, S. P. Rothenberg, J. M. Sequeira, T. Opladen, N. Blau, E. V Quadros, and J. Selhub, “Autoantibodies to folate receptors in the cerebral folate deficiency syndrome.,” N. Engl. J. Med., vol. 352, no. 19, pp. 1985–1991, May 2005. [3] V. T. Ramaekers, J. M. Sequeira, N. Blau, and E. V Quadros, “A milk-free diet downregulates folate receptor autoimmunity in cerebral folate deficiency syndrome.,” Dev. Med. Child Neurol., vol. 50, no. 5, pp. 346–52, May 2008. [4] P. Moretti, S. U. Peters, D. Del Gaudio, T. Sahoo, K. Hyland, T. Bottiglieri, R. J. Hopkin, E. Peach, S. H. Min, D. Goldman, B. Roa, C. a Bacino, and F. Scaglia, “Brief report: autistic symptoms, developmental regression, mental retardation, epilepsy, and dyskinesias in CNS folate deficiency.,” J. Autism Dev. Disord., vol. 38, no. 6, pp. 1170–1177, Jul. 2008.
29
30
ˇarek M. – Significance of Cerebral Folate Deficiency for Development and Progression of Autism Krsiˇcka D., S´
[5] S. U. Steele, S. M. Cheah, A. Veerapandiyan, W. Gallentine, E. C. Smith, and M. A. Mikati, “Electroencephalographic and seizure manifestations in two patients with folate receptor autoimmune antibody-mediated primary cerebral folate deficiency.,” Epilepsy Behav., vol. 24, no. 4, pp. 507–12, Aug. 2012. [6] R. E. Frye, J. M. Sequeira, E. V Quadros, S. J. James, and D. A. Rossignol, “Cerebral folate receptor autoantibodies in autism spectrum disorder.,” Mol. Psychiatry, vol. 18, no. 3, pp. 369–81, Mar. 2013. [7] R. S. Al-Baradie and M. W. Chaudhary, “Diagnosis and management of cerebral folate deficiency. A form of folinic acidresponsive seizures.,” Neurosciences (Riyadh)., vol. 19, no. 4, pp. 312–6, Oct. 2014. [8] D. A. Rossignol and R. E. Frye, “Melatonin in autism spectrum disorders: a systematic review and meta-analysis.,” Dev. Med. Child Neurol., vol. 53, no. 9, pp. 783–92, Sep. 2011. [9] R. Coben, M. Linden, and T. E. Myers, “Neurofeedback for autistic spectrum disorder: a review of the literature.,” Appl. Psychophysiol. Biofeedback, vol. 35, no. 1, pp. 83–105, 2010. [10] K. Bertoglio, S. Jill James, L. Deprey, N. Brule, and R. L. Hendren, “Pilot study of the effect of methyl B12 treatment on behavioral and biomarker measures in children with autism.,” J. Altern. Complement. Med., vol. 16, no. 5, pp. 555–560, May 2010. [11] R. E. Frye, D. Rossignol, M. F. Casanova, G. L. Brown, V. Martin, S. Edelson, R. Coben, J. Lewine, J. C. Slattery, C. Lau, P. Hardy, S. H. Fatemi, T. D. Folsom, D. Macfabe, and J. B. Adams, “A Review of Traditional and Novel Treatments for Seizures in Autism Spectrum Disorder: Findings from a Systematic Review and Expert Panel.,” Front. public Heal., vol. 1, p. 31, 2013.
[12] E. A. Langley, M. Krykbaeva, J. K. Blusztajn, and T. J. Mellott, “High maternal choline consumption during pregnancy and nursing alleviates deficits in social interaction and improves anxiety-like behaviors in the BTBR T+Itpr3tf/J mouse model of autism.,” Behav. Brain Res., Oct. 2014. [13] K. Williams, A. Brignell, M. Randall, N. Silove, and P. Hazell, “Selective serotonin reuptake inhibitors (SSRIs) for autism spectrum disorders (ASD).,” Cochrane database Syst. Rev., vol. 8, p. CD004677, Jan. 2013. [14] V. T. Ramaekers, B. Th¨ ony, J. M. Sequeira, M. Ansseau, P. Philippe, F. Boemer, V. Bours, and E. V Quadros, “Folinic acid treatment for schizophrenia associated with folate receptor autoantibodies.,” Mol. Genet. Metab., Oct. 2014. [15] A. Ho, D. Michelson, G. Aaen, and S. Ashwal, “Cerebral folate deficiency presenting as adolescent catatonic schizophrenia: a case report.,” J. Child Neurol., vol. 25, no. 7, pp. 898–900, 2010. [16] Q. Wang, J. Liu, Y.-P. Liu, X.-Y. Li, Y.-Y. Ma, T.-F. Wu, Y. Ding, J.-Q. Song, Y.-J. Wang, and Y.-L. Yang, “[Methylenetetrahydrofolate reductase deficiency-induced schizophrenia in a school-age boy].,” Zhongguo Dang Dai Er Ke Za Zhi, vol. 16, no. 1, pp. 62–6, Jan. 2014. [17] Z. Sadighi, I. J. Butler, and M. K. Koenig, “Adult-onset cerebral folate deficiency.,” Arch. Neurol., vol. 69, no. 6, pp. 778–779, 2012. [18] H. Blehaut, C. Mircher, A. Ravel, M. Conte, V. de Portzamparc, G. Poret, F. H. de Kermadec, M.-O. Rethore, and F. G. Sturtz, “Effect of leucovorin (folinic acid) on the developmental quotient of children with Down’s syndrome (trisomy 21) and influence of thyroid status.,” PLoS One, vol. 5, no. 1, p. e8394, Jan. 2010.
Semantic Interoperability in Biomedicine and Healthcare
Muˇzn´y M. et al. – Integration of Various Lifestyle and Diabetes Devices . . .
Integration of Various Lifestyle and Diabetes Devices Within a Diabetes Self-management Application Miroslav Muˇ zn´ y1,2 , Martina Vlas´ akov´ a1 Jan Muˇ z´ık1,2 , Eirik Arsand3 1 2 3
First Faculty of Medicine, Charles University in Prague, Czech Republic
Faculty of Biomedical Engineering, Czech Technical University in Prague, Czech Republic
Norwegian Centre for Integrated Care and Telemedicine, University Hospital of North Norway, Tromso, Norway
Correspondence to: Miroslav Muˇ zn´ y Spin-off Application Centre, First Faculty of Medicine, Charles University in Prague Address: Studnickova 7, 128 08 Prague 2 E–mail:
[email protected]
Aims of Research People with diabetes have an opportunity to better manage their disease by using various devices, which may help them understand how their body reacts in different situations. Specifically, this can be facilitated by modern smartphone capabilities and use of a digital diabetes diary, acting as a natural aggregator and interpreter of collected data. An additional important role can be played by wearable devices, which are able to integrate multiple sensors and therefore may effectively merge functionality of multiple single-purpose devices. However, integrating data from various sensors raises the risk of possible personal health data misusage. The sensitive data-flow into a digital diabetes diary has to be handled appropriately carefully to avoid putting their users at risk of health complications [1]. So, what kind of mobile diabetes applications might be relevant to users? Potential applications and sensors are those gathering physical activity data, nutrition data, heart rate data, weight data, data from insulin pumps, and (continuous) blood glucose data. Such data can be supplied by physical activity trackers, smartwatches, smart-scales and body sensors. Each of the elements, which is part of a digital diabetes diary system, may provide a new point of view on personal diabetes self-management. However, due to lack of secure, open communication interface, every device has to be integrated separately, paying attention to possible risks for the patient user. In our research we study the possibilities of utilizing different commercially available diabetes and lifestyle devices within our smartphone application, the Diabetes Diary [2]. Semantic Interoperability in Biomedicine and Healthcare
State of the Art One of the most discussed issues within using diabetes devices at the time being is the impossibility to have control over the users own data. Currently, the most popular approach is to upload collected data to a cloud-based data storage, which serves as a backup media and also acts as a server node for remote synchronization between multiple devices. Thus, a timely question is: are users really interested in having their personal health data stored on a remote server (in the cloud)? Lifestyle and diabetes device manufacturers are generally protecting direct data acquisition from their devices and provide this feature exclusively to their own software applications []. Therefore the possibilities for a local integration, with the users own choice of smartphone-based applications, are limited. Although, this is officially provided for some of the devices (e.g. Misfit activity tracker [3]), for its majority a direct communication interface is unavailable and it’s availability even decreases when crossing from a category of lifestyle devices (e.g. physical activity trackers) to diabetes devices (e.g. blood glucose meters). Also, advantages of having personal data stored in a cloud are unbalanced by an increased risk of possible security issues, which can lead to a full disclosure of personal data in a worst case.
Application in Biomedicine and Healthcare The last couple of months have shown the importance of community-based movements which provide an alternative software platform on the market. One of those is Nightscout, which can be also found under the #WeAreNotWaiting hash tag identifier on the internet
31
32
Muˇzn´y M. et al. – Integration of Various Lifestyle and Diabetes Devices . . .
[4]. This community’s efforts led to a provision of a smartphone-based setup for the Dexcom continuous blood glucose meters (CGM). This setup allows a user to retrieve data from these CGMs using a generally available equipment (USB OTG cable) and to use the blood glucose data for various purposes (e.g. remotely monitoring a child with diabetes’s blood glucose management). Another example is Diabeto, a startup project which aims to design and develop a device for a better integration of various blood glucose meters within mobile applications [5]. Their device will plug into a jack connector of a blood glucose meter and wirelessly transfers blood glucose measurements to a smartphone application. Similar approach would significantly facilitate use of multiple blood glucose meters and allow more secure data integration with a digital diabetes diary. Thanks to these community-initiated innovations, it is significantly easier to utilize various diabetes data into digital diabetes diaries. Even though all of these innovations are not officially supported by devices’ producers and may require a technical knowledge beyond the one of ordinary user, they open new possibilities on how to utilize relevant data in different scenarios of daily life for a better diabetes treatment.
Acknowledgements
Reference: Renard, Eric, et al. ”Closed-loop insulin delivery using a subcutaneous glucose sensor and intraperitoneal insulin delivery feasibility study testing a new model for the artificial pancreas.” Diabetes Care 33.1 (2010): 121-127. SNOMED CT: 261000004 MeSH: not found ICD10: not found
Insulin Pump Definition: Portable or implantable devices for infusion of insulin. Includes open-loop systems which may be patient-operated or controlled by a pre-set program and are designed for constant delivery of small quantities of insulin, increased during food ingestion, and closed-loop systems which deliver quantities of insulin automatically based on an electronic glucose sensor. Synonyms: Insulin Infusion Systems Reference: National Library of Medicine SNOMED CT: 69805005 MeSH: D007332
This paper has been partially supported by the SVV- ICD10: not found 2015-260158 project of Charles University in Prague.
Activity Tracker
Keywords Wearable Technology Definition: Clothing and accessories incorporating computer and advanced electronic technologies. Synonyms: Wearables, Tech togs, Wearable devices Reference: https://en.wikipedia.org/wiki /Wearable_technology
Definition: Activity monitor worn on the upper limbs to measure physical activities. Synonyms: Physical activity tracker, Physical activity sensor Reference: Yang, Che-Chang, and Yeh-Liang Hsu. ”A review of accelerometry-based wearable motion detectors for physical activity monitoring.” Sensors 10.8 (2010): 7772-7788. SNOMED CT: not found
SNOMED CT: not found
MeSH: not found
MeSH: not found
ICD10: not found
ICD10: not found
Remote Patient Monitoring
Definition: A communications network for providing continuous patient monitoring to provide critical. care services from a remote location. Definition: System for an automated insulin delivery which is made up of three components: a subcuta- Synonyms: Remote monitoring neous glucose sensor, the insulin delivery algorithm (running on a laptop computer), and the intraperi- Reference: Rosenfeld, Brian, and Michael Breslow. ”Telecommunications network for remote patient toneal insulin infusion pump. monitoring.” U.S. Patent No. 7,256,708. 14 Aug. Synonyms: Artificial pancreas 2007.
Closed-loop System
Semantic Interoperability in Biomedicine and Healthcare
33
Muˇzn´y M. et al. – Integration of Various Lifestyle and Diabetes Devices . . .
SNOMED CT: not found MeSH: not found ICD10: not found
References [1] KLONOFF, David C. Cybersecurity for Connected Diabetes Devices. Journal of diabetes science and technology, 2015, 1932296815583334.
Semantic Interoperability in Biomedicine and Healthcare
[2] ARSAND, Eirik, SKROVSETH, Stein Olav, JOAKIMSEN, Ragnar Martin, HARTVIGSEN, Gunnar. Design of an Advanced Mobile Diabetes Diary Based on a Prospective 6-month Study Involving People with Type 1 Diabetes. The 6th International Conference on Advanced Technologies and Treatments for Diabetes, February 27. - March 2. 2013, Paris. France. [3] Misfit Scientific Library. https://build.misfit.com/
Available
from:
[4] The Nightscout Project. http://www.nightscout.info
Available
from:
[5] Diabeto. Available from: http://diabe.to/
34
Rak D., Sv´atek V. – Matching Medical Websites to Medical Guidelines through Clinical Vocabularies
Matching Medical Websites to Medical Guidelines through Clinical Vocabularies Duˇsan Rak1 , Vojtˇ ech Sv´ atek2 1 2
First Faculty of Medicine, Charles University in Prague, Prague, Czech Republic
Faculty of Informatics and Statistics, University of Economics, Prague, Czech Republic
Correspondence to: Duˇsan Rak Address: U Z´ atiˇs´ı 545/9, 14700, Praha 4 E–mail:
[email protected]
Aims of Research We propose and test simple approach exploiting MGL content as a benchmark for the assessment of content quality in medical web sites (WS). It is based on the idea that the information content or at least the scope of a medical text is reflected in the domain terminology used. We consider a possible use of this approach in semiautomatic human-based quality verification and various aspects related to its application. Clinical vocabularies are used to discover medical terminology in both groups of texts (MGL and WS). Both sets of terminology are then compared based on extracted data. The WS content quality is assessed firstly based on general content match (i.e. based on concepts or topics discovered) and secondly based on similarity of the particular terminology used in MGLs. Partial goal is to propose and evaluate suitable methods of aggregation of terminology in MGLs so that only one single standard for WS quality assessment might be applied in the end. Important goal is to evaluate the overall applicability of this approach in the process of semiautomatic quality assessment.
State of the Art
by renowned medical societies and based on results of Evidence-Based Medicine (EBM) [2]. MGLs are published in a well structured text form (aiming for medical professionals) and sometimes additionally in highly formalized form such as GLIF [3]. Purpose of this further formalization is to allow automation of MGL content processing it in information systems.
Application in Biomedicine and Healthcare MGLs form important quality benchmark for any particular area of medicine. It is invaluable source of quality information for GP professionals and same for specialists in WS certification agencies. Despite being very well structured and formalized the use of MGLs for fully automatic content assessment is far from possible yet. Such application would require very advanced machine understanding of processed text. On the other hand, simple and straightforward approach based on similarity analysis between the two extracted terminology sets (from MGLs and WS) seems to be feasible and effective in yielding some meaningful quality measures. Additionally, comprehensiveness of content and quality of terminology used or WS similarity groups can be estimated. Beyond these quantitative measures value-added in form of WS annotated by scientific terminology can be offered to certification authority experts.
Modern technology offers a wide array of possibilities for anyone to publish almost any content freely on the Internet. Because of the importance and delicacy of medical information, the quality of such texts provided to general public seems to be a serious issue nowadays. Specialized organizations (such as HON [1]) scan the Internet actively or audit content/form of submitted WS. As more and more WS for medical domain are created the need for automation of this process is obvious. Unfortunately the Acknowledgements only feasible way to approve the adequacy of the medical information content is human verification today. Best practices in medicine are systematically capThis paper has been partially supported by the SVVtured by medical guidelines (MGL), which are provided 2015-260158 project of Charles University in Prague. Semantic Interoperability in Biomedicine and Healthcare
Rak D., Sv´atek V. – Matching Medical Websites to Medical Guidelines through Clinical Vocabularies
Keywords
SNOMED CT: not found
Annotation
MeSH: D019317
ICD10: not found Definition: Annotation means providing base data (text, video, picture, etc.) by descriptive or analytic nota- Similarity tions, tags or other meta-information (see: linguistic annotation). Definition: A concept whereby a set of documents or terms within term lists are assigned a metric based Synonyms: Linguistic annotation, Tagging on the likeness of their meaning/semantic content. Reference: multiple reference Synonyms: Semantic similarity, Semantic closeness, Proximity, Nearness, Relatedness SNOMED CT: not found MeSH: not found ICD10: not found
Reference: http://en.wikipedia.org/wiki /Semantic_similarity SNOMED CT: not found
Concept
MeSH: not found
Definition: Cognitive unit of meaning — an abstract ICD10: not found idea or a mental symbol sometimes defined as a ”unit of knowledge,” built from other units which Unified Medical Language System (UMLS) act as a concept’s characteristics and inferred from specific instances or occurrences. A concept is typi- Definition: A compendium of many controlled vocabucally associated with a corresponding representation laries in the biomedical field. It was created in 1986 in a language or symbology such as a single meaning by NLM. Its purpose is to facilitate the development of a term. of computer systems that behave as if they ”understand” the meaning of the language of biomedicine Synonyms: Entity, Idea and health. To that end, NLM produces and distributes the UMLS Knowledge Sources (databases) Reference: multiple reference, modified from http: and associated software tools (programs) for use //en.wikipedia.org/wiki/Concept and http:// by system developers in building or enhancing elecwww.thefreedictionary.com/concept tronic information systems that create, process, reSNOMED CT: not found trieve, integrate, and/or aggregate biomedical and health data and information, as well as in informatMeSH: not found ics research. Three UMLS Knowledge Sources exist: the Metathesaurus, the Semantic Network, and the ICD10: not found SPECIALIST Lexicon.
Evidence-Based Medicine (EBM) Definition: EBM is the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients. The practice of evidence based medicine means integrating individual clinical expertise with the best available external clinical evidence from systematic research. The strength of scientific evidence of the risks and benefits of treatments (including lack of treatment) and diagnostic tests are taken into account.
Synonyms: UMLS Reference: NML On-Line SNOMED CT: not found MeSH: D017432 ICD10: not found
References [1] Health On the Net Foundation [Internet]. Available from: http://www.hon.ch/
Synonyms: Evidence-based practice (EBP) (a broader term)
[2] Sackett DL. Evidence based medicine: what it is and what it isn’t. BMJ 1996;312:7
Reference: Sackett DL. Evidence based medicine: what it is and what it isn’t. BMJ 1996;312:7
[3] Ohno-Machado L, et al. The GuideLine Interchange Format: A Model for Representing Guidelines. J Am Med Inform Assoc. 1998 Jul-Aug; 5(4): 357–372.
Semantic Interoperability in Biomedicine and Healthcare
35
36
Seidl L., Hanzl´ıˇcek P. – International Communication Protocols for Interoperability in the Czech Republic
International Communication Protocols for Interoperability in the Czech Republic Libor Seidl1 , Petr Hanzl´ıˇ cek2 1
First Medical Faculty, Charles University in Prague, Prague, Czech Republic 2
ˇ Cargo, Prague, Czech Republic CD
Correspondence to: Libor Seidl Centre for eHealth a Telemedicine Adresa: Studniˇ ckova 7, Prague 2, 120 00, Czech Republic E–mail:
[email protected]
Aims of Research
content, its way of use and will enable to identify obstacles when deploying international protocols to the Czech The aim of this research is to identify possibilities and healthcare. to asses barriers in utilization of international communication protocols in the Czech healthcare. Czech healthcare in field of system’s interoperability State of the Art relies namely on a Czech data protocol called DASTA. This protocol was founded in 1997 and has been developed jointly by Czech vendors to fulfil topical needs of System’s architecture description is in general defined data exchange. Cursory reading the protocol specification in ISO/IEC/IEEE 42010 since 2011 superceding original you will not reveal any definition of a trigger event (when IEEE 1471:2000 norm. This norm also places a system data are communicated), you will not find any seman- into an environment, fulfiling a certain role or a mission tics of a transfer (what is the meaning of data transfer from which stakeholders have profits. - record create/delete), you will not find any communication party responsibilities, and finally you will not find Generic Component Model (GCM)[2] describes an arany consequent communication interactions defined. chitecture of a generic information system by proposing Although not presented that way, the DASTA is rather three orthogonal views: a domain perspective, system’s a standardised structure used in fashion of document han- architecture perspective and a view of software developdling than a real communication protocol. International ment process. These three perspectives form a cube diprotocols on the other hand offer message based protocols vided into several sub-cubes. Every atomic sub-cube ad(HL7 v2, HL7 v3) as well as document based specifications dress a particular problem at respective levels of all three (HL7 CDA, CCR and EN 13606). perspectives. The GCM propose to describe a system in With respect to wide deployment of DASTA, a gradual each sub-cube and then provide interconnection between involvement of international standards will bring a need neighboring sub-cubes by appropriate transformations. to run inter-protocol gateway for various protocols. In this process of continuous change various scenarios will Frank Oemig was the first who proposed an ontolnewly appear and change or vanish later requesting high ogy description of HL7 v2 and HL7 v3 communication flexibility of inter-protocol transformations. standards. In his PhD work [3] he involved the GCM Thus the gateway cannot be a hard-coded single- approach. To obain a machine readable mapping he depurpose transformation. Repetition of conformance tests scribed both standards using a Communication Standards after each upgrade will be expensive and will definitely Ontology (CSO)[4], an upper level ontology. HL7 v2 was slow down any transformation process. represented in OWL by automatic transformation of MS We see a solution in a semi or fully automated con- Access HL7 Database, HL7 v3 was represented in OWL figuration of such gateway based on formalised knowledge by transforming a MIF (Message Interchange File Forabout the enviroment, about protocols itself, and about a mat) into OWL. Mapping between both standards was clinical domain in question. facilitated by a domain ontology (patient administration) Ontology description of protocols and mappings be- ACGT (Advanced Clinico-Genomic Trial Ontology). An tween them will enable an assessment of DASTA protocol overview of all ontology composition is shown in Figure 1. Semantic Interoperability in Biomedicine and Healthcare
Seidl L., Hanzl´ıˇcek P. – International Communication Protocols for Interoperability in the Czech Republic
Figure 1: Composition of ontologies after Frank Oemig [3].
Application in Biomedicine and Healthcare
Keywords Communication protocol
Research in this field should identify suitable ontolo- Definition: Application protocol for electronic data exchange in healthcare. gies, extend them and create bridging ontologies so a comprehensive ontological description of integration environ- Synonyms: Communication standard ment, clinical domain as well as involved communication protocols can be composed. Reference: http://skmtglossary.org/search. aspx?term_id=1161&SearchExp=communication% The research should propose a path how various on20protocol tologies in neighboring GCM sub-cubes interconnect into a working unit. Such description will be suitable for an SNOMED CT: not found automatic gateway configuration as well as for an assessMeSH: not found ment of protocol maturity, suitability of integration in particular situation and for identification of deployment ICD10: not found obstacles.
DASTA
An outcome of the applied research can be a technology demonstration of automatically configured inter- Definition: DASTA is regularly updated open standard protocol gateway for a selected use-case. A combination for communication among information systems in of a formalised knowledge and a reasoner can demonstrate Czech healthcare. It covers clinical, laboratory, staa future way of software development of a complex and tistical and administrative domains. It also contains highly dynamic systems. number of codelists including National Codelist of laboratory procedures, Codelist of Clinical Events and various statistical classifications.
Acknowledgements
Synonyms: Data standard
Reference: http://www.dastacr.cz/ This paper has been partially supported by the SVVSNOMED CT: not found 2015-260158 project of Charles University in Prague. Semantic Interoperability in Biomedicine and Healthcare
37
38
Seidl L., Hanzl´ıˇcek P. – International Communication Protocols for Interoperability in the Czech Republic
MeSH: not found
Gateway
ICD10: not found
Definition: Entity, which provides real-time, two-way communications between terminals.
Ontology Definition: Organization of concepts for which a rational argument can be made. Reference: http://skmtglossary.org/search.aspx? term_id=1319&SearchExp=ontology SNOMED CT: not found MeSH: not found ICD10: not found
Domain Definition: A particular area of interest. Reference: http://skmtglossary.org/search.aspx? term_id=1154&SearchExp=domain SNOMED CT: not found MeSH: not found ICD10: not found
Reference: http://skmtglossary.org/search.aspx? term_id=1006&SearchExp=gateway SNOMED CT: not found MeSH: not found ICD10: not found
References [1] Authors of DASTA. DASTA Evolution — Basic Info — DASTA [Internet]. [Cited 5. April 2014]. Available from: http://www.dastacr.cz/info-1.html [2] Blobel B. Application of the Component Paradigm for Analysis and Design of Advanced Health System Architectures. International Journal of Medical Informatics. 18. July 2000;60:281–301. [3] Oemig F. PhD Thesis: Entwicklung einer ontologiebasierten Architektur zur Sicherung semantischer Interoperabilit¨ at zwischen Kommunikationsstandards im Gesundheitswesen [Internet]. [cited 2012 Jul 26]. Available from: http://www.oemig.de/Frank/phd-thesis.htm [4] Oemig F, Blobel B. A Communication Standards Ontology Using Basic Formal Ontologies. Studies in Health Technology and Informatics. 2010:105–13.
Semantic Interoperability in Biomedicine and Healthcare
Schlenker A., Reimer M. – Big Data in Hospital Information Systems in the terms of Security
Big Data in Hospital Information Systems in the terms of Security Anna Schlenker1,2 , Michal Reimer2 1
Institute of Hygiene and Epidemiology, First Faculty of Medicine,
Charles University in Prague and General University Hospital in Prague, Czech Republic 2
Department of Biomedical Informatics, Faculty of Biomedical Engineering, Czech Technical University in Prague, Kladno, Czech Republic
Correspondence to: Anna Schlenker Institute of Hygiene and Epidemiology, First Faculty of Medicine, Charles University in Prague & General University Hospital in Prague Address: Studniˇ ckova 7, 128 00 Prague 2, Czech Republic E–mail:
[email protected]
Aims of Research
State of the Art
In the healthcare sector amount of data has been accumulated over the years. Nowadays, it is still most of the patient’s documentation on paper. Thankfully, many records have been already converted to the electronic form. There is, of course, involved the processing of large volumes of data from electronic health records, which will gradually keep increasing [1]. Big data can be defined using so-called ’4V definition’ [2]:
In the area of hospital information systems the topic of security is very actual. On the one hand, current consulting rooms working without a computer and without the information system are rare, thankfully. On the other hand, many people are becoming interested in, who has access to these systems and the information in it [4]. Today, we put considerable emphasis on the fact that each user need their own login information. Another important thing is to have different editing rights for individual users in the information system (physician, nurse, laboratory technician, radiologist, technical staff, etc.). For users of hospital information systems are organized training sessions where health care professionals are getting familiar with the need for multifactor security of sensitive patient data [4]. The main reason is to make the hospital staff understand that entering the password is not only thing that delays them from work, but also the thing that can protect them. Then teach users that the information system is not only annoying software that do not let them continue their work without filling some fields, but show them that this field can be important [4].
1. V = volume; This means that the volume of data increases exponentially. 2. V = velocity (speed); There are jobs requiring immediate processing of large volumes of data which are continuously generated. A suitable example may be processing data produced by the camera. 3. V = variety (variability); Besides processing the structured data, there are jobs for processing unstructured text, but also various types of multimedia data.
4. V = veracity (credibility); Uncertain credibility of data due to their inconsistency, incompleteness, ambiguity etc. A suitable example can be illustrated Current State of Security in Hospital on the communication on social networks. Information Systems The use of analysis of the big data can improve the quality of health care. In addition to support decisionmaking can analysis of big data also help with riskanalysis or cost-analysis in healthcare. Another possibility is use in developing medical guidelines [1]. Statistical analysis can also help with the better interpretation of the data stored in hospital information systems [3]. Semantic Interoperability in Biomedicine and Healthcare
Currently, the most of the hospital information systems remember on the safety and can set different rights for different users. Unfortunatelly, user who logs into the system is typically verified just by one method, usually by password. This password might not fulfil any of the conditions
39
40
Schlenker A., Reimer M. – Big Data in Hospital Information Systems in the terms of Security
needed for example for password resistant from the dictionary attack. Users should therefore remember, that their password should not consist of the name of the husband/wife, pet or other full-semantic words. Generally, the password should be long enough (at least 8 characters), composed of large and small letters, numbers and special characters. Secure password is the password which the user do not disclose to anyone (not even his/her husband/wife) and do not write it anywhere. For attackers, there is nothing easier than copy the password that is written in user’s notebook [4]. Another mistake in most hospital information systems is that there is usually no automatic logoff. The reason is probably a ’waste of time’ for health care professionals because of constant logging into the information system. Apart of this, users of the information systems should realize that in the case of leaving the computer with logged in information system can cause the lose or change of the sensitive patient data [4].
Figure 1: Keystroke duration and keystroke latency.
To make the data processing easier and clearer, we have extended the application of the second listview (see Figure 2), where we have the opportunity to display the second record. The application has also a sidebar textbox (see Figure 2) where, after marking a selected row in each entry, application calculate parameters of keystroke dynamics of individual user. Application also allows loading a previously scanned records and give us the possibility of back analysis [5].
Application in Biomedicine and Healthcare For the security enhancement in the field od hospital information systems we have designed an application that uses multi-factor authentication which combines the knowledge factor and the biometric factor. As a biometric factor we use a behavioural characterictic called keystroke dynamics. Keystroke dynamics describes exactly how the user types on the keyboard through the scanning when each key was pressed and when it was released. This application can be used in two ways:
Figure 2: Final version of the application for capturing keystroke dynamics.
1. As a application for static verification (username + The application was implemented in C] within the Mipassword + keystroke dynamics when entering usercrosoft Visual Studio Express 2012. Scanned data are exmane and password). ported to the CSV file.
Acknowledgements
2. As a application for continuous verification (verifies if the user who is working with the information sysThis paper has been partially supported by the SVVtem is the same user who is logged on). 2015-260158 project of Charles University in Prague. The biggest advantage of this application is collecting Keywords the data directly from the operating system, without any delay. The application records the key code, key name, Big Data key press time and release time. Definition: Big Data can be defined using so-called ’4V Our application also enables automatic analysis of definition’, where V means volume, velocity, variety scanned data. For the purpose of analysis the applica(variability) and veracity (credibility). tion calculates the timing vector, which consists of the keystroke duration which means the period of time a key Reference: [1] is held for, and the keystroke latency which means the SNOMED CT: not found time between individual keystrokes (see Figure 1). Semantic Interoperability in Biomedicine and Healthcare
Schlenker A., Reimer M. – Big Data in Hospital Information Systems in the terms of Security
MeSH: not found
Reference: [6]
ICD10: not found
SNOMED CT: not found
Information System
MeSH: not found
ICD10: not found Definition: Integrated set of files, procedures, and equipment for the storage, manipulation, and retrieval of Keystroke Dynamics information. Reference: http://www.nlm.nih.gov/cgi/mesh/ 2015/MB_cgi?mode=&term=Information+Systems &field=entry#TreeL01.700.508.300 SNOMED CT: 706593004 MeSH: D007256 ICD10: not found
Hospital Information System
Definition: The detailed timing information that describes exactly when each key was pressed and when it was released as a person is typing at a computer keyboard. Synonyms: Typing Dynamics Reference: [7] SNOMED CT: not found MeSH: not found
Definition: Integrated, computer-assisted systems de- ICD10: not found signed to store, manipulate, and retrieve information concerned with the administrative and clinical References aspects of providing medical services within the hos[1] Raghupathi W., Raghupathi pital. Synonyms: HIS Reference: http://www.nlm.nih.gov/cgi/mesh/ 2015/MB_cgi?mode=&term=Hospital+Information +Systems&field=entry#TreeL01.700.508.300. 408 SNOMED CT: 462944003 MeSH: D006751 ICD10: not found
Multifactor Authentication Definition: A security system in which more than one form of authentication is implemented to verify the legitimacy of a transaction.
Semantic Interoperability in Biomedicine and Healthcare
V.: Big data analytics in healthcare: promise and potential. Health Information Science and Systems 2014, 2(3). doi:10.1186/2047-2501-2-3
[2] Schlenker A., Bohunˇ ca ´k A.: Keystroke Dynamics for Security Enhancement in Hospital Information Systems. International Journal on Biomedicine and Healthcare 2015; 3(1):41–44 [3] Kalina J.: Statistical Challenges of Big Data Analysis in Medicine. International Journal on Biomedicine and Healthcare 2015; 3(1):24–27 [4] Schlenker A.: Multifactor Data Security in Information Systems in Health Care. International Journal on Biomedicine and Healthcare 2014; 2(1):25–27 [5] Reimer M., Schlenker A. Ovˇ eˇrov´ an´ı identity na z´ akladˇ e kontinu´ aln´ıho sn´ım´ an´ı dynamiky stisku poˇ c´ıtaˇ cov´ ych kl´ aves. ˇ Kladno, 2015. Bachelor thesis. Cesk´ e vysok´ e uˇ cen´ı technick´ e. [6] Badr Y., Chbeir R., Abraham A., Hassanien A.-E. (Eds.): Emergent Web Intelligence: Advanced Semantic Technologies, 1st Edition., 2010, XVI, 544, p.345 [7] Bergadano F., Gunetti D., Picardi C.: User authentication through Keystroke Dynamics. ACM Transactions on Information and System Security (TISSEC), 2002;5(4): 367–397
41
42
Siˇcov´a K. et al. – Secondary Cataract in Patients after Implantation of Multifocal IOLs
Secondary Cataract in Patients after Implantation of Multifocal IOLs Krist´ına Siˇ cov´ a1 , Petr V´ yborn´ y1 , Jiˇr´ı Paˇsta1 1
Ophthalmology department, Central military hospital Prague, Czech Republic
Correspondence to: Krist´ına Siˇ cov´ a Ophthalmology department, Central military hospital Prague Address: U Vojensk´ e nemocnice 1200, 16902 Prague 6 E–mail:
[email protected]
Aims of Research With the recent rapid development of refractive eye surgery in developed countries, there are a growing number of operations carried out in patients of middle and old age, who come to see the ophthalmologist wishing to get rid of corrective lenses, or at least dramatically reduce their use. Modern refractive surgery offers this option to patients thanks to a wide range of intraocular lens implants. In the Czech Republic, multifocal diffractive intraocular lenses have recently become the ones most often implanted. This premium procedure demands not only a highquality surgeon and patient’s cooperation, but also requires a supplementary payment from the patient. Therefore, it can be assumed that the patient expects the best possible outcome from this surgery, and any postoperative complications or worsening of the visual acuity becomes a significant problem. The most frequent late complication after an otherwise uncomplicated surgery, is the development of posterior capsule opacification (PCO) of intraocular lenses, decreasing the visual acuity, which can lead to the development of amblyopia. The previously mentioned problem becomes even more important in patients who undergo a premium IOL complex optics implantation surgery. Any complication after such surgery significantly reduces the possibility of using a multifocal lens optics. The aim of our research is to evaluate which factors lead to the development of a secondary cataract, whether its development is affected only by the performance of the surgery, surgical technique used, and the instrumentation, or whether there are other contributing factors such as the age, sex, and comorbidity of the patient, the presence or absence of a cataract preoperatively, the type of cataract, and the extent of preoperative refractive error.
In our clinic, we have been performing refractive surgery with implantation of premium IOLs since 2005, and therefore have got a relatively large amount of followup patients. This fact allows us to evaluate the recorded data both retrospectively and prospectively, using modern objective imaging methods.
State of the Art Currently, it is known that proliferation and migration of epithelial cells (LEC-lens epithelial cells) in particular areas of the lens equator are responsible for the posterior capsule opacification. Secondary cataract pathogenesis has two forms: fibrous changes in the anterior and posterior leaf of the lens capsule, and regeneration in the sense of newly formed lens tissue. Visual acuity is only affected by opacity located in the optical axis of vision. Contact between the optical part of the IOL (intraocular lens) with the lens epithelial cells of the anterior capsule leads to their differentiation into myofibroblasts, followed by the development of fibrosis and ACO (anterior capsule cataract). Their migration to the posterior leaf causes PCO. The emergence of the regenerative type of secondary cataract is caused by the proliferating and migrating LEC, which were left behind during surgery in the equator zone (E-called cells). Firstly, the remaining LEC produce a new lens material along the equator (Sommering ring), and secondly, they (LEC) migrate between the posterior leaf and IOL, where they create the Elschnig pearls [1]. The prevention of secondary cataracts has been shown to be influenced by the shape and material of the IOL, the intraoperative and postoperative treatment, and the surgical technique. The current preventive practice is the IOL implantation into the capsular bag, the use of IOL with sharp edges, and the extensive cleaning of the capsule. Even afSemantic Interoperability in Biomedicine and Healthcare
Siˇcov´a K. et al. – Secondary Cataract in Patients after Implantation of Multifocal IOLs
ter all these interventions, the development of secondary cataracts is not exceptional [2]. After the first post-op year, secondary cataracts affect around 11% of patients (in our study 1,25%) ; five years after the surgery, up to 28% of patients can be affected [3], (in our group 2,34%) (Figure 1.)
Figure 1: Percentage incidence of secondary cataract after implantation MIOL over the years.
These results, independent of the material or the type of the intraocular implant, are in our opinion caused by our system of careful selection of suitable candidates for surgery, the protocol of the operation and postoperative therapy, as well as the post-operative checks, including established strict regime . Therefore, let us say that the results of our research may find an application both in cataract surgery and vitreoretinal surgery, as well as in the health sector when assessing the economic efficiency of procedures. Results of statistical analysis will certainly be of high value to opthalmologists. Moreover, the data about the comorbidity, age, and sex of the patients can offer interesting information to other fields of medicine, such as biomedical statistics. Perhaps the results of this study can help patients decide whether or not to undergo the premium IOL implantation surgery.
Acknowledgements
The average variation is from 10% to 40% in three to five years after the surgery [4]. For some types of IOL it This paper has been partially supported by the SVVis only 5% [5]. 2015-260158 project of Charles University in Prague.
Application in Biomedicine and Healthcare
Keywords Posterior capsule opacification (PCO)
Secondary cataracts, even small-scale ones, can lead to a reduction of visual acuity in patients with multifocal intraocular lens. Worsening of vision manifests itself especially in reading for the short distance. This small PCO, which is in monofocal IOL insignificant, can develop half a year after the surgery [6]. The most common therapy remains NdYAG capsulotomy. There is a low risk of developing tears in the retina, retinal detachment, vitreous opacities formation, displacement or decentration of the IOL, and the deterioration of its function. Despite the high precision measurement of the intraocular lens power, the uncomplicated performance of surgery, and the initially uncomplicated postoperative course, there is a possibility of developing far-reaching complications. The above-mentioned complications may require further surgical intervention, including vitreo-retinal surgery and hospitalization of the patient. This can result in a permanent reduction of visual acuity, patient’s disability, limitations in work and daily life, and thus ensuing financial consequences for both the patient and the healthcare. Due to the already mentioned increasing interest of phacosurgeons and patients in the multifocal implants, the importance of this issue increases. The aim of ophthalmic surgeons is to reduce the risk of secondary cataracts to a minimum, and to maintain good functioning implants for as long as possible. The provisional results of our research, present a lower incidence of secondary cataract than reported in the world literature. Semantic Interoperability in Biomedicine and Healthcare
Definition: Opacitfication of the posterior capsular leaf. Synonyms: Secondary cataract Reference: Kanski J., Bowling B., Nischal K., Pearson A., Clinical Ophthalmology, A systemic approach, 7th edition, 2011; 295 SNOMED CT: 410567004 MeSH: D058442 ICD10: H264
Cataract Definition: Clouding of the natural intraocular crystalline lens that focuses the light entering the eye onto the retina. This cloudiness can cause a decrease in vision and may lead to eventual blindness if left untreated. Reference: Kanski J., Bowling B., Nischal K., Pearson A., Clinical Ophthalmology, A systemic approach, 7th edition, 2011; 295 SNOMED CT: 193570009 MeSH: D002386 ICD10: H25
43
44
Siˇcov´a K. et al. – Secondary Cataract in Patients after Implantation of Multifocal IOLs
Intraocular lens (IOL)
Reference: Kanski J., Bowling B., Nischal K., Pearson A., Clinical Ophthalmology, A systemic approach, Definition: Artificial lens intended for permanent stor7th edition, 2011; 295 age in the eye. Reference: Steinert R.F. Cataract surgery, 3th edition, SNOMED CT: 85785001 2010; 439-442 MeSH: D064727 SNOMED CT: 385468004 MeSH: D007910
ICD10: not found
ICD10: Z961
References Multifocal intraocular lens (MIOL) Definition: Intraocular lens. Reference: Steinert R.F. Cataract surgery, 3th edition, 2010; 439-442 SNOMED CT: 385468004 SNOMED CT: 313236002 MeSH: not found ICD10: Z961
NdYAG capsulotomy Definition: Surgical incision of the posterior lens capsule to remove the secondary cataract neodymium laser.
[1] Sacu S., Menapace R., Findl O. et al. Influence of optic Eege design and anterior capsule polishing on posterior capsule fibrosis. Journal of cataract and refractive surgery, 2004; 30: 658-662 [2] Krajˇ cov´ a P., Chynoransk´ y M., Strmeˇ n P. Opacifik´ acia zadn´ eho p´ uzdra ˇsoˇsovky po implant´ acii rˆ oznych typov umel´ ych vn´ utrooˇ cn´ ych ˇsoˇsoviek - II. ˇ casˇt : rˆ ozne peroperaˇ cn´ e n´ alezy. ˇ Cesk´ a a slovensk´ a oftalmologie, 2008; 64: 13-15 [3] Bertelmann E., Kojetinsky C. Posterior capsule opacification. Current opinion in Ophthalmology, 2001; 12: 35-40 [4] Pandey S.K., Apple D.J., Wener L. et al. Posterior capsule opacification: A review of ethiopathogenesis. Experimental and clinical studies and factors for prevention. Ophthalmology, 2004;52 : 99-112 [5] Hayashi K., Hayashi H., Posterior capsule opacification in the presence of an intraocular lens with a sharp versus rounded optic edge, Ophthalmology, 2005; 112 : 1550-1556 [6] Larkin H., PCO and premium lens, Eurotimes, 2012; 17(5): 4-6
Semantic Interoperability in Biomedicine and Healthcare
45
Slov´ak D., Zv´arov´a J. – DNA in Biomedical Applications
DNA in Biomedical Applications Dalibor Slov´ ak1,2 , Jana Zv´ arov´ a1 1
Institute of Hygiene and Epidemiology, First Faculty of Medicine, Charles University in Prague, Czech Republic 2
Institute of Health Information and Statistics of the Czech Republic, Prague, Czech Republic
Correspondence to: Dalibor Slov´ ak Institute of Health Information and Statistics of the Czech Republic Address: Palack´ eho n´ am. 4, 128 01 Prague 2 E–mail:
[email protected]
Aims of Research The use of DNA is now commonly accepted practice in the professional as well as non-professional public. It is used in many areas of criminology: the identification of the perpetrators of the crime, in paternity testing, in identification of victims of mass disasters and other cases. With the decreasing price of analyzes lately expand use of DNA analysis in many other ways: examination of historical remains, compiling family trees and searching and kinship ties, and also in biomedical aplications.
mined. For physicians it is important to perform a test before starting treatment, or when considering the extension or change of treatment. One example is the VKORC1 gene, which has 24 known alleles. Two of them (CYP2C92, CYP2C9-3) enhance the anticoagulant effect of warfarin and reduce the daily doses needed to maintain an International Normalized Ratio (INR) in the therapeutic range. In the UGT1A1 gene is for carriers of allele 28 observed severe toxicity of Irinotecan, a drug used mainly to treatment of metastatic tumors of the colon and rectum.
State of the Art
Application in Biomedicine and Healthcare
The first example of use in medical sciences is Huntington’s chorea. Expansion in the number of trinucleotide repeat sequences in Huntington’s chorea increases the probability of an unbalanced mental state. In serious crimes, it is possible to assess the sanity of the individual and his ability to control. As the state outside the norm is indicated the number of repetitive repetitions above 35. In the Czech population there is an innate tendency to increased blood coagulation (thrombophilia) relatively highly extended – affects about 8 % of the population. Most risk group are women taking normal hormonal contraception which have genetic predisposition to increased coagulation. To prevent and minimize the problems associated with the development of thrombosis and thromboembolic disease it is important to know the genetic disposition, because an endangered person may appropriately adjust their diet and lifestyle on the basis of their knowledge of hereditary dispositions. Therefore, there is an analysis of thrombophilic mutations also in donors of gamets. The most common is factor V Leiden (G1691A), causing resistance of Factor V to anticoagulant activity of APC protein. One of the motivations for the development of personalized medicine is that the effect of some drugs can be affected by the genetic information of the patient, i.e. response of organism to drugs can be genetically deter-
DNA has two big advantages for testing. Human DNA consists of approximately 23,000 genes, some of which have yet more transcripts so characters usable for differentiation are even more. Identical entire genome have no two people in the world, not even identical twins. Second advantage of DNA is its omnipresence. DNA is present in every cell and remains in all places where we move. However, this advantage of easy availability can be a disadvantage at the same time. Due to the sensitivity of current techniques it is possible to analyze even a few cells. In analyzing such samples there are often found mixed DNA samples or contamination by impurities. In proportion to this there increases the importance of including all possible stochastic phenomena associated with laboratory processing to the calculation. Another problem of DNA analysis is a restriction on the selected part of genome. In forensic practice there is not evaluated the entire DNA, but so called DNA profile is compiled. One reason is the price: a human genome sequencing has cost less than 1 million dollars and is continuing to fall, but for routine analysis it is still a very high price. The second reason is the sensitivity of the information that can be obtained from DNA. Therefore, in more complex cases clear identification is not possible and it is necessary to use probabilistic models, including the
Semantic Interoperability in Biomedicine and Healthcare
46
Slov´ak D., Zv´arov´a J. – DNA in Biomedical Applications
various number of parameters. These models usually in some way calculate the probability of observing this DNA profile in the population. Because of the simplicity, some used models can ignore the parameters that may be crucial for the result, or may include these parameters to the calculation in a wrong way. From this there rise the mistakes that can fundamentally change the entire result of DNA analysis. As a memento, it is possible to note a case described in the article [1]: The authors took a mixed sample of DNA evidence from a real crime and presented it to 17 experienced analysts working in accredited government lab in the USA, without any contextual information that might bias their judgement. They reached opposing conclusions about whether the suspect matched the sample or not: only one analyst agreed with the original judgement that the suspect ”cannot be excluded”. Four analysts said the evidence was inconclusive and 12 said he could be excluded. Subsequent survey of labs around the world also shows that there are significant inconsistencies in the guidelines on how to interpret a sample. As we can see, DNA analysis also has its risks. It is necessary to unify the terminology, improve and standardize probabilistic models and have respect to the guidelines. However, even then it will not be possible to completely eliminate the stochastic character of certain phenomena. Therefore, it is necessary to see DNA analysis as important, but not infallible tool of modern science.
Synonyms: Nucleic acid repetitive sequence
Acknowledgements
SNOMED CT: 234467004
Reference: http://aboutforensics.co.uk/dna-analysis/ SNOMED CT: 62302004 MeSH: not found ICD10: not found
Repetitive Sequence Definition: Multiply repeated sequence e.g. nucleotides in the DNA; they represent a large portion of human genome.
Reference: www.lekarske.slovniky.cz SNOMED CT: 51512005 MeSH: D012091 ICD10: not found
Thrombophilia Definition: Increased tendency to formation of blood clots. Reference: www.lekarske.slovniky.cz
This paper has been partially supported by the SVV- MeSH: D019851 2015-260158 project of Charles University in Prague. ICD10: D68.5
Keywords Deoxyribonucleic Acid
International Normalized Ratio Definition: The blood test used to assess correct dosage of anti-clotting drugs.
Definition: A type of nucleic acid which is the basis of genetic information. Synonyms: INR, Standardized prothrombin time Synonyms: DNA
Reference: www.urmc.rochester.edu/encyclopedia
Reference: www.lekarske.slovniky.cz SNOMED CT: 24851008
SNOMED CT: 165581004
MeSH: D004247
MeSH: D019934
ICD10: not found
ICD10: not found
DNA Analysis Definition: Procedure of forensic science that uses biological material containing DNA.
References [1] Geddes L. Fallible DNA evidence can mean prison or freedom. New Sci. 2010; 207 (2773): 8-11
Semantic Interoperability in Biomedicine and Healthcare
47
Stonov´a M. – Unstructured Data in Evidence-based Healthcare
Unstructured Data in Evidence-based Healthcare Michaela Stonov´ a1 1
First Faculty of Medicine, Charles University, Prague, Czech Republic
Correspondence to: Michaela Stonov´ a First Faculty of Medicine, Charles University, Prague, Czech Republic Address: Kateˇrinsk´ a 32, 121 08 Prague E–mail:
[email protected]
Aims of Research The Holy grail of unstructured data analysis is an Automatic Diagnostic System (ADS). To achieve this ultimate goal a lot of consecutive steps must be done first. Understanding a Natural Language Processing (NLP) is one of them. During the NLP customisation to medical and local needs a by-product for a medical Big Data analysis was created[3]. Evidence-based Healthcare (EBH) and its subpart Evidence-base Medicine (EBM)[1] are heavily dependent on the ability to correctly assess vast amount of data. Narrative text representing 80 % of Electronic Medical Records (EMR) is a valuable source, yet still omitted due to its unstructured form. Acquired knowledge and skills from the NLP adjustment permitted to build a system for a Narrative Medical Records (NMR) analysis.
State of the Art The NLP customisation is a real corner stone for analysis of unstructured data. Not only the language morphology, syntax and semantics must be incorporated, but also a high entropy medical jargon must be considered[7]. A prebuilt model was tested on 386 587 EMRs from a Hospital Information System (HIS) of the Central Military Hospital Prague (CMHP). The testing records were obtained from various outpatient departments (e.g. emergency, cardiology, dermatology, ophthalmology, and so on). Patients records extracted directly from HIS databases were anonymized. Physician’s and patient’s names were erased and the National Identification Number was replaced by unique number, in this case the Patient’s Number (PN).Thus privacy of all patients was protected, but the possibility to link relevant EMRs to a certain patient was maintained. The final format of each record was UTF-8 encoded text file following this semistructured form: 1 Based
on ICD-10 code [4]. name for this facet is Medical Product database
2 Correct
facet [6].
Semantic Interoperability in Biomedicine and Healthcare
age;gender;patient;diagnosis1 ;narrative text. 29;Male;675032;H353 ;Patient suffered . . . A median file size was 727 B. The file sizes oscillated between few bytes (the smallest file was only 26 B long containing just one word) and several kilobytes (the biggest over 15,3 kB) in a testing set. A total sum of all these testing 386 587 text files was 1,6 GB. The medical records were processed in the Apache Lucene based system IBM Watson Content analytics. All input data were standardly crawled, parsed and indexed [5]. For a farther medical analysis these five special ad hoc categories (called facets) were created: -
Age facet, Gender facet, Patient facet, Diagnosis facet, Drug facet.
The Age facet contains sub-facets for each life decade, two special sub-facets for under aged and teen aged patient (both due to the fact, that CMHP is dedicated for adult patients) and one misfit category for new-borns and typing errors. The Gender facet distinguishes between male and female patients. The Patient facet shows all EMRs of a picked patient (based on the Patient’s Number). The Diagnosis facet divides disease into 22 main groups and 276 subgroups according to its ICD-10 code [4]. The last Drug facet 2 represents the Anatomical Therapeutic Chemical (ATC) classification system, where the active substances are divided into different groups according to the organ or system on which they act and theirs therapeutic, pharmacological and chemical properties. The 5 862 sub-facets copy a drugs division into fourteen main groups (1st level), each with subsequent subgroups until the possible 5th level. Based on the last two facets, the system is currently able to recognize and catalogue 38 617 different diagnosis 3 The original number of registered items by the State institute for Drug Controll was almost 56 000. The majority of drugs is produced by several pharmaceutical companies under the same name, so it was possible to lower the origin number to 4 256.
48
Stonov´a M. – Unstructured Data in Evidence-based Healthcare
based on ICD-10 code[4] and 4 2563 medicinal products registered by the State institute for Drug Control, including e. g. the newest one in the L01X Antineoplastic and Immunomodulating Agent ATC group4 . During the processing, each word in the input text file is not only linguistically recognised, but also labelled if it belongs to a certain facet. All collateral information about each word are than stored in the Index, in our case in 69 files of total size 4,8 GB. Five added facets and standard NLP processing enriched the Index thus far, that indexed data are triple the origin size (1,6 GB). Once indexed data are easily visualised in the IBM Watson Content Analytics user interface. One click on corresponding facet reveals that from 386 587 records, 211 586 patients were men and 175 001 were woman. The majority of them (58 026 patients) have visited the outpatient department repeatedly, the rest of them only once (35 898 patients). In total 93 924 different outpatients have come to the CMHP and were treated for 4 525 different diagnosis (according to its ICD-10 code). The previous results are also possible to obtain from a standard database due to the structured form of the first four parts of each EMR. Finding how many smoking patients were treated by other antineoplastic agents (L01X ATC group) for which diagnosis, is not the case. The information about smoking habits is generally stored in the narrative text, as a medication. Performing the content data analysis on unstructured data is the only feasible option. The basic Czech NLP allows to filter not only words in a base form as kuˇr´ ak 5 or kuˇraˇcka 6 or kouˇrit 7 , but 8 also their inflexions . This NLP content analysis revealed 13 484 smoking patients. Fifteen of them were treated by L01X ATC group drugs for various diagnosis (Table 1).
Only 244 EMRs were labelled as alcohol-attributable disease (ICD-10 codes F100 and F102), but other 549 records explicitly mentioned that a patient was drunk in the narrative part of EMR. In total at least 793 patients were intoxicated (Table 2)9 .
Table 2: Inebriated patients.
The proposed system is designed for quick analysis10 of unstructured data. System is able to deal with Big Data (up to several terabytes of data). After data crawling and indexing, the analysis can be performed instantly. The only limitation is medical language entropy. In some cases proposed system distorts the analysis. Prime example is shown in both tables. Scientific notations x10 in lab results (e.g. Thrombocytes: 176.00 x10ˆ9/l) coincide11 with the ICD-10 code X10 for a Contact with hot drinks, food, fats and cooking oils diagnosis. High number of X10 diagnosis in both cases is erroneous, but in the overall statistic is negligible. This type of distortion is more probable in short abbreviations, but also longer names like a ketoconazole can cause some misinterpretations. The ketoconazole can be both, an antifungal agent for topical use (D01AC08 ATC group) and a hormone suppressant (H02C ATC group). Which case is relevant depends on the context. Despite the fact, that mentioned faults have a minor impact on the statistic in the Big Data scope, the next step of research will be dedicated to its mitigation. Further research will focus on the content analysis improvement and a spelling error correction.
Application in Biomedicine and Healthcare The proposed analytic system is able to easily process terabytes of narrative medical records and is independent of data format. Therefore it can be used on various data The number of patients coming to the policlinic in an sources without further customisation. The system is preinebriated state can be other task for the content analysis. pared for an automatic classification based on age, gender, Table 1: Smoking patients.
4 Especially
monoclonal antibodies and various inhibitors. smoker. 6 A female smoker. 7 To smoke. 8 E.g. kouˇ r´ı, kouˇ ril, kouˇ rila, kuˇ r´ ack´ a, kuˇ r´ akem, kuˇ r´ ack´ y and so on. 5A
9 Table
2 shows only the diagnosis with at least ten occurrences. perform both smoking and drinking analysis took two minutes, including a results export to a csv file. 11 The system is deliberately case insensitive. 10 To
Semantic Interoperability in Biomedicine and Healthcare
49
Stonov´a M. – Unstructured Data in Evidence-based Healthcare
diagnosis, and used medication. With acceptable accu- SNOMED CT: not found racy, the system is prepared for a survey Big Data analysis of EMRs and EHRs. The limitation based on medical MeSH: D009323 language entropy will be addressed in the next steps. ICD10: not found
Acknowledgements
Structured Data
This paper has been partially supported by the SVV- Definition: Data that resides in fixed fields within a record or file. 2015-260158 project of Charles University in Prague.
Keywords Big Data
Synonyms: Structured Information Reference: Webopedia http://www.webopedia.com/ TERM/S/structured_data.html
SNOMED CT: not found Definition: Big data is high volume, high velocity, and/or high variety information assets that require MeSH: not found new forms of processing to enable enhanced decision making, insight discovery and process optimisation. ICD10: not found Reference: Gartner http://www.gartner.com/ Unstructured Data it-glossary/big-data Definition: Refers to information that either does not SNOMED CT: not found have a pre-defined data model or is not organized in a pre-defined manner. MeSH: not found ICD10: not found
Reference: The Free Dictionary http://encyclopedia. thefreedictionary.com/Unstructured+data
Evidence-Based Medicine
SNOMED CT: not found
Definition: The Conscientious, explicit and judicious use MeSH: not found of current best evidence in making decisions about ICD10: not found the care of individual patients. Synonyms: Safe Medicine Reference: Cochrane collaboration http://www.cochrane.org/ SNOMED CT: not found MeSH: D019317 ICD10: not found
Natural Language Processing Definition: The branch of information science that deals with natural language information. Synonyms: Human Language Technology, NLP Reference: The Free Dictionary http://www.thefreedictionary.com/natural+ language+processing
Semantic Interoperability in Biomedicine and Healthcare
References [1] Cochrane Collaboration, https://www.cochrane.org.
Available
from:
[2] Hartzband P, Groopman J. Untangling the Web –Patients, Doctors, and the Internet. N Engl J Med 2010; 362:1063–1066. [3] Holzinger A, Stocker C, Ofner B, Prochaska G, Brabenetz A, Hofmann-Wellenhof R. Combining HCI, Natural Language Processing, and Knowledge Discovery –Potential of IBM Content Analytics as an Assistive Technology in the Biomedical Field. Human-Computer Interaction and Knowledge Discovery in Complex, Unstructured, Big Data.2013; 7947: 13–24. [4] Institute of Health Information and Statistics of the Czech Republic, Available from: https://www.uzis.cz. [5] Stonov´ a M. Unstructured Data in Healthcare. Semantic Interoperability in Biomedicine and Healthcare. IJBH 2014; 2(1): 34–36. [6] State Institute for https://www.sukl.eu.
Drug
Control,
Available
from:
[7] Zvolsk´ y M. Automating the Use of Clinical Practice Guidelines in the Health Information Infrastructure. Semantic Interoperability in Biomedicine and Healthcare. IJBH 2014; 2(1): 51–52.
50
ˇ nkov´a B., Beneˇs J., Sedmera D. – Conduction System Development in Mouse Saˇ
Conduction System Development in Mouse ˇ nkov´ Barbora Saˇ a1,2 , Jiˇr´ı Beneˇs1,2 , David Sedmera1,2 1 2
Institute of Anatomy, First Faculty of Medicine, Charles University, Prague, Czech Republic Institute of Physiology, Academy of Sciences of the Czech Republic, Prague, Czech Republic
Correspondence to: ˇ nkov´ Barbora Saˇ a Institute of Anatomy, First Faculty of Medicine Address: U Nemocnice 3, 128 08 Prague 2, Czech Republic E–mail:
[email protected]
Aims of Research Function and Development of Cardiac Conduction System Cardiac conduction system is responsible for generation and spread of the electrical impulse in the heart. The intrinsic rhythm of the heart is determined by the sinoatrial node located in the right atrium near the entry of the superior caval vein. Electrical impulse is subsequently propagated through the fast conducting atrial myocardium to reach the atrioventricular node at the border of the atria and ventricles which functions as a part of mechanism of generating a momentary delay in propagation of action potential and it enables efficient ventricular filling with blood during a diastole. Following exit from the atrioventricular node, action potential rapidly propagates along the His bundle and its distal branches, finally activating the ventricular myocardium by a highly ramified network of the Purkinje fibers. The larger fascicles of ventricular conduction system are insulated from the surrounding myocardium as they course from the crest of the septum toward ventricular apex. This insulation breaks down within the peripheral network of Purkinje fibers, enabling direct electrical coupling with the working myocardium. This arrangement causes the activation of ventricular myocardium from the apex of the heart and from the endocardium to epicardium, resulting in a characteristic electrocardiogram. Each element of cardiac conduction system is differentiated during development in a distinctive pattern that appears to be morphogenetically conserved. This morphogenetic sequence is conserved in chick, mouse and humans [4]. Pacemaker is the first functional compartment of the conduction system to be recognized in the primitive tubular heart, in mouse it appears around embryonic day (ED) 7.5. Cardiac impulse propagates in a slow isotropic fashion from the sinus venosus located caudally through the primitive ventricle towards cranially located outflow tract
[5]. Slowly conducting atrioventricular canal appears in the looping heart (in mouse ED8), separating the faster conducting regions of the atria and ventricles. Slowly conducting segments (atrioventricular canal, outflow tract and sinoatrial region) function as a sphincter-like valves with coordinated contraction to increase efficiency of the early blood pump. Ventricular activation sequence in the looped heart follows blood flow. Primitive activation pattern “primary interventricular ring” utilizing preferential activation along the future interventricular septum [11] emerges in the developmental window before septation of the heart (in mouse ED9.5-12.5). Fast conducting ventricular conduction system is the last element of cardiac conduction system to differentiate and its function is accompanied by an apparent reversal in the ventricular activation sequence from an immature base-to-apex towards a mature sequence from apex-to-base. The mature apexto-base activation coincides with completion of ventricular septation in the chick [3, 8] whereas in the mouse it appears earlier in the development, before interventricular septum is finished [9].
The function of cardiac conduction system in the developing heart is best assessed using the high speed optical mapping, the main technique employed in this study. Optical mapping employs a system of virtual electrodes and it is very convenient method for the millimeter size embryonic hearts, it is based on usage of voltage dependent die and signal recording by high speed camera. Optical mapping provides in vivo physiological data presented by epicardial spatio-temporal activation maps, result of optical mapping recoding analysis. First activated site on the epicardial surface and the direction of electrical impulse propagation can be evaluated from activation maps to make together a pattern of the ventricular activation, marker of maturity of conduction system. Another parameter which can be evaluated from activation maps is speed of electrical impulse propagation in the epicardium. Semantic Interoperability in Biomedicine and Healthcare
ˇ nkov´a B., Beneˇs J., Sedmera D. – Conduction System Development in Mouse Saˇ
Normal Development of Mouse Conduction System Morphology of the conduction system and its development in the mouse were described histologically in the 1970s, and recently, optical mapping has provided clues about its functional deployment. However, quantitative evaluation of ventricular activation times and patterns, necessary for interpretation of changes observed in transgenic mouse models with potential impact on the conduction system development, have still been lacking. Applying method of optical mapping of action potential propagation, we have studied function of conduction system in mouse embryos starting at 9th embryonic day. We measured total activation time of the left ventricle and evaluated way of ventricular activation by classification into activation patterns; namely activation utilizing primary ring, left bundle branch (LBB), right bundle branch (RBB), both bundle braches and transitional types. Typical situation at any given ED is represented by a spectrum of several activation patterns, therefore, it is impossible to describe normal development with analysis of less than 10 embryos per group. The first primitive activation pattern to appear in the chick embryos is activation from the base to apex, in contrast with situation during mouse conduction system development, this type of activation was never observed in the mouse embryos. Early conduction at ED9.5-11.5 is characteristic by utilizing the primary ring, structure in the future interventricular septum. As development proceeds, activation through the primary ring disappears, and at ED13.5 no heart was activated by this structure. LBB is the first active ventricular conduction system compound but with appearance of transitional type of activation. RBB or both branches become active at ED11.5. At ED14.5 the majority of hearts is activated from two centers, but occasionally originated from single one. Spread of electrical impulse is accelerated in the developmental window ED10.5-12.5, as evidenced by shortening of left ventricular epicardial activation time. Thereafter, the activation time remains flat but since the heart continues to grow, the apparent speed of conduction increases. Quantification of normal development of ventricular activation provides a necessary framework for analysis and of transgenic mouse with suspected developmental conduction system defects. For interpretation of changes observed in mouse models is necessary systematic, quantitave study of mouse activation patterns. Data describing normal development were used as a background in studies analyzing mouse lacking Cx40 [11, 2], Tbx2, Ptx2 [1], Erbb2 genes and in mouse model for Long QT Syndrome [10].
Cx40 Deficiency
mains unchanged whereas in the ventricles becomes restricted to trabeculae, resulting in expression in the ventricular conduction system tissue (bundle branches and Purkinje fiber network) in adult. Cx40 deficiency causes embryonic and atrial conduction anomalies [7] and slower conduction with right bundle branch block in the ventricular conduction system [13, 15, 6]. However, the functional importance of Cx40 in the developing conduction system has not been studied. Atrial conduction velocity heterogeneity, dependent on Cx40 genotype, was observed in ED12.5 and ED14.5 embryos. In wild type and heterozygous embryos the first activated site was in the area of presumptive sinoatrial node with moderately prolonged epicardial activation time in heterozygotes (significant ED12.5). Two atrial activation patterns appeared in knockout mice, activation from sinoatrial node and predominate ectopic activation originating in right atrial appendage. Strong prolongation of activation time in Cx40 deficient mice was more prominent in ectopic activated hearts and in ED12.5 compared to ED14.5. Lack of Cx40 in atria during development influences the activation patterns of impulse propagation and significantly slows impulse propagation velocity (which directly correlates with the type of atrial activation). No differences among genotypes were obtained during analysis of ventricular activation times. Frequency of activation patterns revealed notable decrease in the frequency of active LBB at ED12.5 and 14.5 in Cx40-deficient mice, with heterozygotes showing an intermediate phenotype. The proportion of activation patterns began to reverse at ED16.5, where functional LBB was present in nearly 100% in all genotypes, while the frequency RBB activation pattern began to decrease in the Cx40-deficient hearts, suggesting a developing RBB dysfunction. These differences were even more prominent at ED18.5, where Cx40deficient hearts presented only 33% of functional RBB in contrast to 96% in heterozygotes and 94% in wild-type hearts.
State of the Art The study has been completed and data published [11, 1, 10, 2]. Optical mapping and confocal microscopy whole mount imaging represent cutting edge techniques for studying physiological and morphological properties of the embryonic tissues. Both of these are now well established at the First Faculty of Medicine.
Application in Biomedicine and Healthcare
Understanding signaling mechanism involved in conduction system development may be of significance to Cx40 is the main gap junction protein expressed in clinicians and basic researchers studying adult cardiac atria and ventricles in the early phases of embrygenesis. diseases. Congenital abnormalities together with ectopic As embryogenesis proceeds Cx40 expression in atria re- or inappropriate induction of conduction tissues in adult Semantic Interoperability in Biomedicine and Healthcare
51
52
ˇ nkov´a B., Beneˇs J., Sedmera D. – Conduction System Development in Mouse Saˇ
heart are processes that contribute to arrhythmias in adults. Arrhythmias represent a group of cardiovascular diseases that significantly contributes to morbidity and mortality in the population.
Acknowledgements This study has been partially supported by the SVV2015-260158 project of Charles University in Prague, MSMT VZ 0021620806, AS CR AVOZ50450515, GACR 304/08/0615.
Keywords Optical Mapping Definition: Voltage-Sensitive Dye Imaging Synonyms: Action Potential Optical Mapping Reference: [12] SNOMED CT: not found MeSH: D056969 ICD10: not found
Right Bundle Branch Block Definition: A form of heart block in which the electrical stimulation of heart ventricles is interrupted Synonyms: Bundle-Branch Block Reference: [14] SNOMED CT: 404684003 MeSH: D002037 ICD10: I45.1
Connexin 40 Definition: Gap junction protein Synonyms: Gja5 protein Reference: [6] SNOMED CT: not found MeSH: C082919 ICD10: not found
References [1] Ammirabile G, Tessari A, Pignataro V, Szumska D, Sardo FS, Benes J, Jr., Balistreri M, Bhattacharya S, Sedmera D, Campione M. 2012. Pitx2 confers left morphological, molecular, and functional identity to the sinus venosus myocardium. Cardiovasc Res 93:291-301. [2] Benes J Jr, Ammirabile G, Sankova B, Campione M, Krejci E, Kvasilova A, Sedmera D. 2014. The role of connexin40 in developing atrial conduction. FEBS Lett. 2014 Apr 17;588(8):1465-9. [3] Chuck ET, Freeman DM, Watanabe M, Rosenbaum DS. 1997. Changing activation sequence in the embryonic chick heart. Implication for the development of the His-Purkinje system. Circ Res 81:470-476. [4] Gourdie RG, Harris BS, Bond J, Justus CHKW, O’Brien TX, Thompson RP, Sedmera D. 2003. Development of the Cardiac Pacemaking and Conduction System. Birth Defects Research 69:46-57. [5] Kamino K. 1991. Optical approaches to ontogeny of electrical activity and related functional organization during early heart development. Physiol Rev 71:53-91. [6] Kirchhoff S, Nelles E, Hagendorff A, Kruger O, Traub O, Willecke K. Reduced cardiac conduction velocity and predisposition to arrhythmias in connexin40-deficient mice. Curr Biol 1998;8:299–302. [7] Leaf DE, Feig JE, Vasquez C, Riva PL, Yu C, Lader JM et al. Connexin40 imparts conductionheterogeneity to atrial tissue. Circ Res 2008;103:1001–1008. [8] Reckova M, Rosengarten C, deAlmeida A, Stanley CP, Wessels A, Gourdie RG, Thompson RP, Sedmera D. 2003. Hemodynamic is a key epigenetic factor in development of the cardiac conduction system. Circ Res 93:77-85. [9] Rentschler S, Vaidya DM, Tamaddon H, Degenhardt K, Sassoon D, Morley GE, Janife J, Fishmann GI. 2001. Visualization and functional characterization of the developing murine cardiac conduction system. Development 128:1785-1792. [10] de la Rosa AJ, Dominguez JN, Sedmera D, Sankova B, HoveMadsen L, Franco D, Aranega A. 2013. Functional suppression of Kcnq1 leads to early sodium channel remodeling and cardiac conduction system dysmorphogenesis. Cardiovasc Res. 2013 Jun 1;98(3):504-14. [11] Sankova B, Benes JJ, Krejci E, Dupays L, Thevenian-Ruissy M, Miquerol L, Sedmera D. 2012. The effect of connexin40 deficiency on ventricular conduction system function during development. Cardivas Res:doi:10.1093/cvr/cvs1210. [12] Sedmera D, Reckova M, Rosengarten C, Torres MI, Gourdie RG, Thompson R P. 2005. Optical Mapping of Electrical Activation in the Developing Heart. Micros Microanal 11: 209-215. [13] Simon AM, Goodenough DA, Paul DL. Mice lacking connexin40 have cardiac conduction abnormalities characteristic of atrioventricular block and bundle branch block. Curr Biol 1998;8:295–298. [14] Stejfa M et al. 2006. Kardiologie. Grada. 776p. [15] Tamaddon HS, Vaidya D, Simon AM, Paul DL, Jalife J, Morley GE. High-resolution optical mapping of the right bundle branch in connexin40 knockout mice reveals slow conduction in the specialized conduction system. Circ Res 2000;87:929–936.
Semantic Interoperability in Biomedicine and Healthcare
ˇ Steffl M., Plz´ak J. – Presence of Nasal Microbiota and Their Influence on Development of Chronic Rhinosinusitis
Presence of Nasal Microbiota and Their Influence on Development of Chronic Rhinosinusitis 1 ˇ Martin Steffl , Jan Plz´ ak1 1
Department of Otorhinolaryngology and Head and Neck Surgery, 1st Faculty of Medicine Charles University in Prague, Faculty Hospital Motol
Correspondence to: ˇ Martin Steffl Department of Otorhinolaryngology and Head and Neck Surgery, 1st Faculty of Medicine Charles University in Prague, Faculty Hospital Motol ´ Address: V Uvalu 84, 150 06 Prague 5, Czech Republic E–mail:
[email protected]
Aims of Research The aim of this research is to analyze microbial communities, which are presented in nasal a paranasal sinus, their comparison in patients with chronic rhinosinusitis with or without polyps to health controls (patients without any diseases of sinonasal tract). Therefore the scientific question is: ”What is the role of changed nasal microbiota in development chronic rhinosinusitis, its subtypes and if any correlation to clinical findings can be seen?” Many of the past studies were done using the conventional, cultivation based methods. These have many limitations, especially for anaerobes and the data are often unrepresentative. Therefore the recent research is focused on detection of microbiota using modern, cultureindependent techniques [2]. Bacterial DNA obtained from sinonasal region using nasal swabs and biopsies will be amplified using specific primers for 16S DNA. Finally, sequencing analysis will be done. The clinical and paraclinical examination of the patients will be performed. We will correlate these data to the laboratory results. Comparison to the other studies focused on this issues could be another interesting output. Also dissimilarity according to the used methods could be interesting.
State of the Art
nosinusitis with the presence of nasal polyps and chronic rhinosinusitis without nasal polyps [3]. Chronic rhinosinusitis is a disease causing the decrease of work productivity and quality of life. Clinical symptoms of the disease are mostly nasal obstruction, headaches and loss of smell. Management of chronic rhinosinusitis forms yearly significant amount of costs. Treatment is just symptomatic, often tricky and needs repeated surgical revisions [3]. Presence of different microbial communities in nasal and paranasal sinuses was demonstrated in the past studies. The exact connection between the involvement of chronic rhinosinusitis and the presence of these microbes was not demonstrate. Factors like genetical changes, allergies, infections, mucous tissue remodeling, aberrant immunomodulation cause chronic inflammation, which is the template for formation of chronic rhinosinusitis [5]. The infiltration of mucosa with eosinophils, neutrophils, lymphocytes and macrophages, aberrant regulation of Th1 and Th2 lymphocytes, defects in mucous, specific and nonspecific immunity, damage of epithelia followed by aberrant remodeling, eicosanoid changes, fibrosis and swelling are all then present. The mechanism, which causes all these changes is still unknown [4]. Also the role of bacteria is unknown. According to the past studies such an influence is predicted, but interpretation and comparison of these studies is quite difficult due the differences in their methodics. In spite of this differences, different microbiology in acute and chronic rhinosinusitis was found [3].
Chronic rhinosinusitis is a very common disease (515% of population in Europe and USA) affecting the upper respiratory tract. Unfortunately the exact etiology of the disease is still unclear. It is probably multifactorial. Application in Biomedicine and The role of microbiota, especially of bacterial communities is predicted, but not exactly proved [1]. It is characterised Healthcare as chronic inflammation localised in nasal and paranasal sinuses, which persists for at least 12 weeks. Chronic rhiThe outcome of this study should be findings, if the nosinusitis could be divided into two groups – chronic rhi- differences in microbiome in patients with chronic rhinosSemantic Interoperability in Biomedicine and Healthcare
53
54
ˇ Steffl M., Plz´ak J. – Presence of Nasal Microbiota and Their Influence on Development of Chronic Rhinosinusitis
inusitis and persons without any symptoms of sinonasal Sequencing of Nucleic Acids disorder are present and if yes, what the role of this microbiota in formation of chronic rhinosinusitis is. We intend Definition: Process of determining the precise order of to to answer the question, if the new molecular techniques nucleotides within a DNA molecule. could be applied in management and treatment of chronic Synonyms: molecular-genetic analyses rhinosinusitis.
Acknowledgements
Reference: https://en.wikipedia.org/wiki/DNA_ sequencing
SNOMED CT: 117040002 This paper has been partially supported by the SVV2015-260158 project of Charles University in Prague. MeSH: D008969
Keywords Chronic Rhinosinusitis Definition: Sinonasal inflammation lasting for more than 12 weeks.
ICD10: not found
Microbiome Definition: Collection of genomes of microbes in a system. Synonyms: Microbiota
Synonyms: Rhinitis, sinusitis Reference: www.hopkinsmedicine.org/ otolaryngology/specialty_areas/sinus_ center/conditions/sinusitis.html
Reference: microbe.net/2015/04/08/what-does -the-term-microbiome-mean-and-where-did -it-come-from-a-bit-of-a-surprise/ SNOMED CT: not found
SNOMED CT: 40055000
MeSH: D064307
MeSH: D012220, D012852
ICD10: not found
ICD10: J31.0, J32
Nasal Polyps Definition: Soft, painless, noncancerous growths on the lining of nasal passages or sinuses. Synonyms: Pouch of the nasal mucosa Reference: www.mayoclinic.org/ diseases-conditions/nasal-polyps/basics/ definition/con-20023206 SNOMED CT: 52756005 MeSH: D009298 ICD10: J33
References [1] Frank, D., Feazel, L., Bessesen, M., Price, C., Janoff, E., & Pace, N. (2010). The Human Nasal Microbiota and Staphylococcus aureus Carriage. PLoS ONE. [2] Stressmann, F., Rogers, G., Chan, S., Howarth, P., Harries, P., Bruce, K., & Salib, R. (2011). Characterization of bacterial community diversity in chronic rhinosinusitis infections using novel culture-independent techniques. Am J Rhinol Allergy American Journal of Rhinology and Allergy, 133-140. [3] Kennedy, D. (2012). Rhinology diseases of the nose, sinuses, and skull base. New York: Thieme. [4] Kato, A. (n.d.). Immunopathology of chronic rhinosinusitis. Allergology International, 121-130. [5] Li, C., Shi, L., Yan, Y., Gordon, B., Gordon, W., & Wang, D. (2012). Gene Expression Signatures: A New Approach to Understanding the Pathophysiology of Chronic Rhinosinusitis. Curr Allergy Asthma Rep Current Allergy and Asthma Reports, 209-217.
Semantic Interoperability in Biomedicine and Healthcare
55
Vesel´a Fl´orov´a Z. et al. – Glaucoma Treatment for 1 CZK per Day – Dream or Reality?
Glaucoma Treatment for 1 CZK per Day – Dream or Reality? Zuzana Vesel´ a Fl´ orov´ a 1 , Petr V´ yborn´ y 2 , Silvia Siˇ c´ akov´ a 2 , Jiˇr´ı Obenberger 1 2 3
3
Eye Clinic JL Prague, Czech Republic
Eye Clinic 1st Faculty of Medicine and Central Military Hospital – Military University Hospital in Prague, Czech Republic
Department of Radiology of Faculty Hospital Bulovka and the 1st Medical Faculty of Charles University in Prague, Czech Republic
Correspondence to: Zuzana Vesel´ a Fl´ orov´ a Eye Clinic JL Prague, Ltd. Address: V H˚ urk´ ach 1296/10, 158 00 Prague 5, Czech Republic E–mail:
[email protected]
Aims of Research
prostaglandin analogues 5 CZK (5.0807CZK) and a group of medicines in principle, therapeutically interchangeable Recently we have encountered problems with a signif- medicinal products containing fixed combination of timicant reduction of prescription for fluorinated quinolones olol and prostaglandin analogs has set the base payment ”E, OPH” in ophthalmic practice [3]. Constraints origi- for UDTD 5.5537CZK [1]. nating from the administrative decision of the State Institute for Drug Control (SIDC) surprised the experts. The Application in Biomedicine and situation that occurred, motivated us to become more interested in the processes involved in the determination of Healthcare reimbursement of drugs. [4] How the conservative treatment of a patient with glauSIDC considers that the results of the evaluation of coma will specifically draw on the health insurance system cost effectiveness made in other countries can not be con? How the eye drops are reimbursed? How much spends verted to a health environment in the Czech Republic. It the state on the treatment of this serious disease by indi- is necessary in case of adaptation of foreign assesment to vidual therapeutic groups? take into account these important factors: the character of the medical practice, the intensity of resources, the level of costs, the definition of the target population and other State of the Art key assumptions specific to usual clinical practice in the Websites of state institutions - in this case the Ministry Czech Republic [1]. Administrative proceedings usually take several of Health and SIDC are freely accessible to professionals months [2]. And what about the results? Annual savand laymen [1, 2]. ings in millions CZK are expected. These materials can be used as a relevant source of Based on the reimbursement of the groups of prodinformation in the field of drug policy. There are numucts in principle therapeutically interchangeable in therbers of details from the course of negotiations on setting apy of glaucoma, SIDC estimates that the implication payments for the respective treatment groups. It is posof restrictions on the public health insurance will be sible to read multi-year chronology of the process of decost savings as follows: for the treatment with betatermining the remuneration, objections and suggestions blockers approx. 16 621 570 CZK, for the treatment with of pharmaceutical companies that are trying to prioritize prostaglandins approx. 60 938 769 CZK, and finally for their product and get a percentage bonus for it, because the combo-treatment prostaglandin/timolol approx. 41 some publications speak in its favor. The other publica623 642 CZK. tions speak in favor of other companies, and are therefore The estimate was prepared on the basis of consumplogically given the competition [1, 2]. tion of products in the year 2012, with the prices valid on From these materials it is possible to learn a lot about January 5th, 2013 [1, 2]. the evaluative mechanism of SIDC’s objections, and what Estimated total savings achieved because of the three is the basis for a final decision on the payment amount. administrative proceedings is almost 120 millions CZK. Specific result is alarming – the usual daily therapeutic dose (UDTD) for beta-blockers has fixed basic What is the planned utilization of this amount? Is it payment of less than 1CZK (0.9412CZK), UDTD for possible that a refund from health insurance system deSemantic Interoperability in Biomedicine and Healthcare
56
Vesel´a Fl´ orov´a Z. et al. – Glaucoma Treatment for 1 CZK per Day – Dream or Reality?
termined for daily glaucoma treatment using topical betablockers [3] was valued less than the lowest value coin in contemporary monetary system in the Czech Republic? And what happens in the future? What will be the impact of further SIDC’s revisions? Will we have in our country the lowest level of reimbursement of medical treatment in the European system? What about the portfolio of eye drops to treat glaucoma in a few years? Do new medicines ever come in such a condition of low payment? Will be the cost acceptable for pharmaceutical companies to start a registration of a new medical product? How much is a patient able and/or willig to pay at the pharmacy? Are some of the eye drops with unacceptable supplements ”practically negotiable”? Will be under the current conditions the holder of the registration of the drug interested in its extension? How many drugs will be no longer available? Will be the antiglaucoma drugs distributed via generic companies only? Continual efforts to save cost of medicines covered by health insurance system could implicite serious negative consequences, the restriction of the therapeutic portfolio for both – the patiens and the physicians, with an important socio-economic impact. [1] It should be noted that the the practical result of these limitations will be fully operational with a latency of several years. That is the important reason to start the discussion of this issue now.
MeSH: C11.525.381 ICD10: H40-H42
Drug
Definition: A drug is, in the broadest of terms, a chemical substance that has known biological effects on humans or other animals. Foods are generally excluded from this definition, in spite of their physiological effects on animal species. In pharmacology, a drug is ”a chemical substance used in the treatment, cure, prevention, or diagnosis of disease or used to otherwise enhance physical or mental well-being.” Pharmaceutical drugs may be used for a limited duration, or on a regular basis for chronic disorders. Psychoactive drugs are chemical substances that affect the function of the nervous system, altering perception, mood orconsciousness. Alcohol, nicotine, and caffeine are the most widely consumed psychoactive drugs worldwide. Recreational drugs are drugs that are not used for medicinal purposes, but are instead used for pleasure. Common recreational drugs include alcohol, nicotine and caffeine, as well as other substances such as opiates and amphetamines. Some drugs can cause addiction and habituation and Acknowledgements all drugs can cause side effects. Many drugs are illegal for recreational purposes and international This paper has been partially supported by the SVVtreaties such as the Single Convention on Narcotic 2015-260158 project of Charles University in Prague. Drugs exist for the purpose of legally prohibiting certain substances.
Keywords Glaucoma Definition: Glaucoma is a term describing a group of ocular (eye) disorders that result in optic nerve damage, often associated with increased fluid pressure in the eye (intraocular pressure) (IOP). The disorders can be roughly divided into two main categories: ”open-angle” and ”closed-angle” (or ”angle closure”) glaucoma. Open-angle chronic glaucoma is painless, tends to develop slowly over time and often has no symptoms until the disease has progressed significantly. It is treated with either glaucoma medication to lower the pressure, or with various pressure-reducing glaucoma surgeries. Closedangle glaucoma, however, is characterized by sudden eye pain, redness, nausea and vomiting, and other symptoms resulting from a sudden spike in intraocular pressure, and is treated as a medical emergency.
Reference: https://en.wikipedia.org/wiki/Drug SNOMED CT: 410942007 MeSH: not found ICD10: not found
Refund Med from Health Insurance Definition: The government defines the scope of care covered by public health insurance based on medical criteria, degree of disability and the scope of public health insurance. Also defines the temporal and local access to healthcare and saves taxpayers such availability for the insured to provide. Reference: http://www.aktualne.cz/wiki/politika/ zdravotnictvi/r~i:wiki:740/ SNOMED CT: not found
Reference: https://en.wikipedia.org/wiki /Glaucoma
MeSH: not found
SNOMED CT: 23986001
ICD10: not found Semantic Interoperability in Biomedicine and Healthcare
Vesel´a Fl´orov´a Z. et al. – Glaucoma Treatment for 1 CZK per Day – Dream or Reality?
SIDC
Reference: http://www.sukl.eu/sukl/historyand-present
Definition: The State Institute for Drug Control is an administration body established by the Act no. 79/1997 Coll. It falls under direct control of the Ministry of Health. The scope of operation of the Institute is stipulated by legal regulations. In order to safeguard its tasks, the Institute establishes regional workplaces located outside the headquarters of the Institute. The Institute’s mission is, in the interest of public health protection,
SNOMED CT: not found MeSH: not found ICD10: not found
Antiglaucoma Definition: A medicine used to treat glaucoma.
Reference: http://www.olecich.cz/slovnik/ • to ensure that all human pharmaceuticals availantiglaukomatikum able on the Czech market meet appropriate standards of quality, safety and efficacy SNOMED CT: 419886007 • to take share in ensuring that only safe and functional medical devices are used in the MeSH: not found Czech Republic, in addition, accompanied by reliable and appropriate information. More- ICD10: not found over, its role is to contribute to rational use and where appropriate, to responsible and ethReferences ical clinical trials of both medicinal products and medical devices. [1] http://www.mzcr.cz, Data used on December 29th, 2014 • Regulatory procedures shall not result in un- [2] http.//www.sukl.cz, Data available under the numbers of SIDC’s decisions necessary obstacles to availability of medicinal products and medical devices nor to introduc- [3] V´yborn´y P., Siˇc´akov´a S., Dohnalov´a P., Feˇrtek M., Doleˇzal T.: Treatment of glaucoma – a current overview of data and infortion of new therapeutic procedures. In the mation. Czech and Slovak Ophthalmology 69, 3, 2013, 118-126. Czech Republic the Institute ensures the following act [4] http://www.prolekare.cz/glaukom-novinky/
Semantic Interoperability in Biomedicine and Healthcare
57
58
Vlas´akov´a M. et al. – Telemonitoring of the Basic Therapeutic Elements of the Diabetes Mellitus Treatment . . .
Telemonitoring of the Basic Therapeutic Elements of the Diabetes Mellitus Treatment and Their Evaluation Martina Vlas´ akov´ a1 , Miroslav Muˇ zn´ y1,2 , Jan Muˇ z´ık3 1 2
The First Faculty of Medicine, Charles University in Prague, Prague, Czech Republic
Faculty of Biomedical Engineering, Czech Technical University in Prague, Kladno, Czech Republic 3
Spin-off company and research results commercialization center, Prague, Czech Republic
Correspondence to: Martina Vlas´ akov´ a The First Faculty of Medicine Address: Katerinska 32, 121 08 Prague 2 E–mail:
[email protected]
Aims of Research The main goal of the research is to make a technical solution for better and more effective health service for diabetes patients. The modern technology can improve the metabolic control of the disease and make better quality of patient’s life. This goal will be achieved using the telemedicine system (Figure 1) which integrates three main technologies: the modern smart phone application which allows recording of many parameters, mainly carbohydrate intake, blood glucose values, physical activity and insulin doses, the activity tracker for online monitoring of physical activity and the glucometer for measuring and online sharing of the blood glucose level. All data from the glucometer, activity tracker and mobile phone app are shared with a web server using a smart phone in the real-time. The system allows online data transmission and synchronization from continuous monitoring system. The mobile application can be loaded to the intelligent smart watch too. This technical solution allows two-way data sharing with a smart phone.
State of the Art Diabetes mellitus, generally called all chronical metabolism diseases, affects especially carbohydrates. Many factors have an influence on the disease compensation. The most important are the amount of carbohydrate intake, insulin bolus and level of physical activity [1]. These parameters can be controlled by many technical devices available in the market. We can get more information and more useful data using these tools. However, there is a problem with assessment of the enormous amount of data and their heterogeneity. Another challenge is how to motivate patients enough for measuring these parameters as it is painful, discomfort, technically
demanding or time consuming and patients may have insufficient education to self-management. But successful disease compensation depends on the high-quality data and its quick evaluation. The telemonitoring is more and more common tool in compensation of diabetes. Doctors get accurate and reliable data in the real-time using the telemonitoring [2]. The remote monitoring can influent stances and behavior of patients and it can potentially improve their state of health [3]. The telemonitoring has an incentive and education effect. It allows quickly and easily evaluate treatment results to the doctor. Telemedicine system brings higher quality of health service and simultaneously it decreases the need to use the health service. It can have a positive effect on treatment costs. The stable health states of patients have an affirmative influence on their overall satisfaction and their living standards. The telemonitoring is based on the communication between transmitter and receiver in the real-time when the immediate reaction of the medical staff is allowed on the basis of the patient’s incentive [4]. The patient uses the interactive device for control his biological parameters which shares data with his doctor via internet. Then the doctor evaluates the data and choses subsequent medical procedures.
Application in Biomedicine and Healthcare The technical solution for online data collection, sharing and centralization of measured parameters of each patient gives the patients and their doctors the clearly and organized record of monitored values. The Telemedicine system allows to record trends in glucose levels and enables to monitor the dependency rate of glucose level on physical activity. A constituent of the system is preproSemantic Interoperability in Biomedicine and Healthcare
Vlas´akov´a M. et al. – Telemonitoring of the Basic Therapeutic Elements of the Diabetes Mellitus Treatment . . .
Figure 1: The Designed Telemedicine system.
cessing and sorting out of the data and ability to perform the statistical comparison of measured values. The positive effect of the system for the patients is a direct feedback, which reflects an influence of their current life style on the disease. Simultaneously the users get better control and supervision of their disease, which potentially increase the patient motivation to keep the set treatment. The benefit for the doctor is a direct and quick supervision of the patient. The doctor might have more time to adjust the treatment. The designed solution contributes to decreasing the negative impact of incorrectly compensated disease (hypoglycemic events due to low levels of blood glucose and comorbidities caused by inadequate compensations of high blood glucose levels). Using of the telemedicine system can lead to reduction of the costs and to increasing the quality of patient life. Semantic Interoperability in Biomedicine and Healthcare
Acknowledgements This paper has been partially supported by the SVV2015-260158 project of Charles University in Prague.
Keywords Diabetes Mellitus Definition: A group of metabolic diseases in which a person has high blood sugar, either because the pancreas does not produce enough insulin, or because cells do not respond to the insulin that is produced. Reference: Velky lekarsky slovnik. http://lekarske. slovniky.cz (accessed 28 June 2014)
59
60
Vlas´akov´a M. et al. – Telemonitoring of the Basic Therapeutic Elements of the Diabetes Mellitus Treatment . . .
SNOMED CT: C0011849 MeSH: D003920 ICD10: E10-E14
Telemedicine
Synonyms: Ambulatory continuous glucose monitoring of interstitial tissue fluid Reference: Edelsberger T. Encyklopedie pro diabetiky, 1st ed. Praha: Maxdorf; 2009. SNOMED CT: 439926003
Definition: Delivery of health services via remote MeSH: not found telecommunications. This includes interactive consultative and diagnostic services. ICD10: not found Synonyms: eHealth, mHealth, Telehealth
Blood Glucose Level
Reference: National Library of Medicine - Medical Subject Headings. http://www.nlm.nih.gov (accessed Definition: Explanation of the word, including the ref26 June 2015) erence, possible synonyms and indexing. SNOMED CT: 448337001 Synonyms: Blood Sugar MeSH: D017216 ICD10: not found
Mobile Applications
Reference: Velk´ y l´ekaˇrsk´ y slovn´ık. http://lekarske. slovniky.cz (accessed 25 June 2015) SNOMED CT: 365812005
Definition: Computer programs or software installed on MeSH: D001786 mobile electronic devices which support a wide range of functions and uses which include television, tele- ICD10: R73 phone, video, music, word processing, and Internet service. Synonyms: Mobile Apps,Portable Electronic Applications Reference: National Library of Medicine - Medical Subject Headings. http://www.nlm.nih.gov (accessed 26 June 2015) SNOMED CT: not found MeSH: D063731
References
[1] Jaana, M. Home telemonitoring of patients with diabetes: a systematic assessment of observed effects. Journal of Evaluation in Clinical Practice 2006; (13): 242–253. http://www.ncbi.nlm.nih.gov/pubmed/17378871 (accessed 10. December 2012). [2] Mignerat M, Lapointe L, Vedel I. Using telecare for diabetic patients: A mixed systematic review. Health Policy and Technology. 2014;3(2):90-112.
Continuous Glucose Monitoring
[3] Glasgow R.E. D net diabetes self management program: long-term implematation, outcomes, and generatization results. Preventive medicine 2003; (36): 410-419. http://www.ncbi.nlm.nih.gov/pubmed/12649049 (accessed 10. December 2012).
Definition: It defines a measurement of blood sugar (glucose) in short intervals (5 minutes) during one or more days by a special sensor placed in the subcutaneous area. Data are transmitted either directly on the display or transferred to a computer.
[4] Stone R.A. The Diabetes Telemonitoring Study Extension: an exploratory randomized comparison of alternative interventions to maintain glycemic control after withdrawal of diabetes home telemonitoring. Journal of the American Medical Informatics Association 2012; (66): 973–979. http://www.ncbi.nlm.nih.gov/pubmed/22610495 (accessed 5. December 2012).
ICD10: not found
Semantic Interoperability in Biomedicine and Healthcare
Vondruˇskov´a L. – Advantages of Virtual Patient in Paramedical fields of the Health Care Services
Advantages of Virtual Patient in Paramedical fields of the Health Care Services Lenka Vondruˇskov´ a1 1
University Hospital in Pilsen, Czech Republic, Neurosurgery Clinic – ICU
Correspondence to: Lenka Vondruˇskov´ a Neurosurgery Clinic - ICU, University Hospital in Pilsen, Czech Republic Address: Dr. E. Beneˇse 13, 301 00 Pilsen E–mail:
[email protected]
Aims of Research
the development and evaluation methods that will lead to an increased use of virtual clinical scenarios. [4] The advantages of virtual clinical scenarios include a reduced risk, in contrast to a real patient. This matter has been dealt with by authors such as Eagles, who points out the disadvantages of using of a real patient [5]; or Zary, who mentions that mistakes are allowed in VP, unlike with a real patient. [6] The advantage of VP with regards to the reduced risk towards real patients is also dealt with by Gordon and Stevens [7, 4]. To ensure the quality of a created VP it is possible to follow the standards of the International Organization for Standardization (ISO). [8, 9] Quality is also dealt with by the European Committee for Standardization (CEN). [10] The development and support of technology standards is the concern of the non-profit international group MedBuiquitous. Consideration should be given to the development of new forms of VP and the use of existing educational support programs, taking into account the financial savings.
Virtual patient (VP) is a term which is not clearly understood. Some imagine a sophisticated, artificially created simulated patient, while others understand it as computer software. The Association of American Medical Colleges defines VP as a specific type of computer program that simulates real-life clinical scenarios; learners emulate the roles of health care providers to obtain history, conduct a physical exam, and make diagnostic and therapeutic decisions. [1] Cook and Triola describe VP as clinical scenarios that are played out on a computer screen. Students assess the patient by estimating, or selecting possible responses with the option of adding, for example, laboratory test results. The computer offers students the answers or additional information about the condition of the patient. Students are expected to set the diagnosis and treatment plan. VP should be intended and used to support clinical decisionmaking skills. [2] According to Hurst the purpose of VP is to educate students and healthcare professionals with the use of computers which simulate the real situation of a healthcare en- Application in Biomedicine and vironment using a virtual instructor and mediation feed- Healthcare back. [3] The aim of this research is to create computerized clinInterest in supporting the development of educational ical scenarios for non-medical healthcare staff and the sub- support for healthcare professionals and the possibility of sequent testing of the effectiveness of these scenarios in using virtual clinical scenarios in non-medical healthcare education. fields is covered by eHealth documents (eHealth Action Plan, i2010 European initiative for growth and employment, eHealth Action Plan (eHAP) 2012–2020). State of the Art Interest is also evident from the Czech Ministry of ˇ in the project Deepening and Increasing We can presume, if the virtual clinical scenario is an Health (MZ CR) image of a real clinical situation and a real choice of di- Levels of Expertise (for Medical and Non-Medical Staff). Non-medical healthcare professionals have a legally esagnostic and therapeutic steps, VP can be a useful supporting tool for testing knowledge and decision-making. tablished duty to educate themselves according to § 67 Stevens points out that the future use of VP depends on of Act no. 96/2004 Coll., Conditions for the AcquisiSemantic Interoperability in Biomedicine and Healthcare
61
62
Vondruˇskov´a L. – Advantages of Virtual Patient in Paramedical fields of the Health Care Services
tion and Recognition of Qualifications for Pursuing Non- Simulation Medical Professions and Activities Related to the Provision of Health Care and Amendments to Certain Related Definition: Fakery of individual events or behaviour of Acts (Non-Medical Professions Act), as amended. the entire System. Two questions are connected to this: can VP reduce Reference: Nˇemeˇcek, M. a kol. Struˇcn´ y slovn´ık didakstaff turnover in healthcare? Can it help to speed up the tick´e techniky a uˇcebn´ıch pom˚ ucek. Praha: St´atn´ı training of new employees in healthcare? pedagogick´e nakladatelstv´ı, 1985, p. 82
Acknowledgements
SNOMED CT: not found
MeSH: D003198 This paper has been partially supported by the SVVICD10: not found 2015-260158 project of Charles University in Prague.
Keywords
Clinical
Definition: Having to do with the clinic. Antithesis can be theoretical (surgery is clinical department, anatomy is department theoretical) or ambulant (clinical and out patients department of hospital. Definition: Interactive computer simulation of a real clinic scenario for the purpose of medical training, Reference: Vokurka, M., Hugo, J. a kol. Praktick´ y education or treatment. slovn´ık medic´ıny. Praha: Maxdorf, 2011, str. 234
Virtual patient
Reference: Ellaway, R., Candler, C., Greene, P., Smoth- SNOMED CT: 58147004 ers, V., 2006. An Architectural Model forMedBiqMeSH: D015510 uitous VPs. MedBiquitous, Baltimore, MD ICD10: Z006 SNOMED CT: not found MeSH: not found ICD10: not found
Risk Definition: Danger, risk, exposure to danger of health, hazard, the probability of suffering harm.
Education
Reference: Topilov´a, V. Anglicko-ˇcesk´ y, ˇcesko-anglick´ y L´ekaˇrsk´ y slovn´ık. Praha: Grada Publishing, 1999, str. 555 Definition: A process organised and realised in special education facilities along with the process of indiSNOMED CT: 129839007 vidual activity. It concerns the acquisition of knowledge, skills, attitudes and development of ability to MeSH: D012306 use this knowledge, skills and attitudes in performance, behaviour and other self-education and the ICD10: not found education of others. Education always has its aims, concrete content and gives the individual a number References of functions in life and society: socialising, individ- [1] Association of American Medical Colleges, 2007. Effective Use ual development, economic, instrumental, generally of Educational Technology in Medical Education: Summary Report of the 2006 AMC Colloquium on Educational Technolcultural, emancipation. The result of the education ogy. AMC, Washington DC process is education. Synonyms: Learning Reference: Kol´aˇr, Z. a kol. V´ ykladov´ y slovn´ık z pedagogiky. Praha: Grada, 2012. ISBN: 978-80-2473710-2. p. 179 SNOMED CT: 2760311006 MeSH: Q000193 ICD10: Z71.9
[2] Cook, D.A., Triola, M.M., 2009. VPs: a critical literature review and proposed next steps. Medical Education 43, 303e311 [3] Hurst, M.H., Marks-Maran, D. Using a virtual patient aktivity to teach nurse prescribing. Nurse Education in Practice 2011; 11:192-198 [4] Stevens, A., Hernandez, J., Johnsen, K., Dickerson, R., Raij, A., Harrison, C., DiPietro, M., Allen, B., Ferdig, R., Foti, S., Jackson, J., Shin, M., Cendan, J., Watson, R., Duerson, M., Lok, B., Cohen, M., Wagner, P., Lind, D.S., 2006. The use of VPs to teach medical students history taking and communication skills. American Journal of Surgery 191, 806e811 [5] Eagles, J., Calder, S., Nicoll, K., Sclare, P.D., 2001. Using simulated patients in education about alcohol misuse. Academic Medicine 76 (4), 395
Semantic Interoperability in Biomedicine and Healthcare
Vondruˇskov´a L. – Advantages of Virtual Patient in Paramedical fields of the Health Care Services
[6] Zary, N., Johnson, G., Boberg, J., Fors, U., 2006. Development, implementation and pilot evaluation of a web-based VP case simulation environment e web-SP. BioMed Central Medical Education 6,10. At. http://www.pubmedcentral.nih.gov/ articlerender.fcgi?tool-pubmed&pubmedid-16504041 (accessed 10.07.09) [7] Gordon, J.A., Wilkerson, W.M., Schafer, D.W., Armstrong, E.G., 2001. Practicing medicine without risk: students’ and educators’ responses to high-fidelity patient simulation. Academic Medicine 76 (5), 469e472
Semantic Interoperability in Biomedicine and Healthcare
[8] ISO/IEC 19796-1. Information technology – Learning, education and training – Quality management, assurance and metrics: Part 1: General approach. Switzerland: ISO/IEC 2005 [9] ISO/IEC 19796-3. Information technology – Learning, education and training – Quality management, assurance and metrics: Part 3: Reference methods and metrics. Switzerland: ISO copyright office, 2009 [10] CEN/ISSS CWA 14644: Quality Assurance and Guidelines. Brussels: CEN/ISSS, 2003
63
64
Zv´ara K. et al. – Preprocessing of Narrative Medical Reports for Information Extraction
Preprocessing of Narrative Medical Reports for Information Extraction Karel Zv´ ara1 , Marie Tomeˇ ckov´ a2 , Vojtˇ ech Sv´ atek3 , Jana Zv´ arov´ a1 1
Institute of Hygiene and Epidemiology, 1st Faculty of Medicine, Charles University in Prague, Czech Republic. 2 3
EuroMISE Mentor Association, Prague, Czech Republic
Department of Information and Knowledge Engineering, Faculty of Informatics and Statistics, University of Economics, Prague, Czech Republic
Correspondence to: Karel Zv´ ara Institute of Hygiene and Epidemiology, 1st Faculty of Medicine, Charles University in Prague Address: Studniˇ ckova 7, 121 08 Prague 2, Czech Republic E–mail:
[email protected]
Aims of Research
epSOS Patient Summary [4]. The primary anticipated task is to provide easy cross-border access to patient’s history, allergies, medications and similar useful information. Other uses include measuring efficiency and quality of healthcare and the possible use of advisory expert systems.
Information extraction from narrative medical reports is an important step for a quality and efficiency of medical decision making. The main research objective is to develop the method for preprocessing of narrative medical reports and to transform information from narrative medical reports to a structured form. The structured information can be further stored in electronic health record A system for processing narrative reports has been cresystems. ated. This system is accessible using web browser and with the following mode of operation. In the first phase user (a physician) selects clinical part of the narrative State of the Art medical report. This part is automatically tokenized and To acquire medical information from a narrative med- prepared for the second phase. The aim of the second ical report fully reusable in electronic healthcare we need phase is to standardize input the narrative medical reto store it in a structured form. Many scientists coped port. It contains several tools to change tokenized inwith this problem. I tis possible to see that the process of put to transform it to the standardized form. The stanextracting information from narrative reports depends not dardized form should not contain abbreviations, can cononly on original language [1] but also on local standards tain commonly used acronyms and must not contain any (incl. legislation) and the subject who creates the docu- mistypes and other typing errors. Completing the secmentation. The ambiguity of terms used has been empha- ond phase lets the user to proceed to the third phase. sized by [2] stating by referencing other sources that half The aim of the third phase is to identify vocabulary items of commonly used abbreviations in ear, nose and throat and match each such item to one of supported codebooks. surgery were unclear to more than 90% of junior doctors Supported codebooks are SNOMED CT, LOINC, ICD10 ´ from other specialties. Some success extracting numeric and database of pharmaceuticals of SUKL. values (e.g. blood pressure) has been achieved, see [3].
Application in Biomedicine and Healthcare
Acknowledgements
The structured information form narrative medical reports can be stored in electronic health record systems and This paper has been partially supported by the SVVfurther used for medical decision making. The structured information should create some kind of useful extract like 2015-260158 project of Charles University in Prague. Semantic Interoperability in Biomedicine and Healthcare
Zv´ara K. et al. – Preprocessing of Narrative Medical Reports for Information Extraction
Keywords
Quality of Care
Report
Definition: The levels of excellence which characterize the health service or health care provided based on accepted standards of quality. Definition: Detailed account or statement or formal record of data resulting from empirical inquiry. Reference: http://www.nlm.nih.gov/cgi/mesh/ Reference: http://www.nlm.nih.gov/cgi/mesh/ 2015/MB_cgi 2015/MB_cgi SNOMED CT: 86078004 SNOMED CT: 229059009 MeSH: D058028
MeSH: D011787
ICD10: not found
ICD10: not found
Legislation Definition: Works consisting of the text of proposed or enacted legislation that may be in the form of bills, laws, statutes, ordinances, or government regulations. Reference: http://www.nlm.nih.gov/cgi/mesh/ 2015/MB_cgi SNOMED CT: not found MeSH: D020485 ICD10: not found
Semantic Interoperability in Biomedicine and Healthcare
References [1] Garcia-Remesal M., Maojo V., Billhardt H., Crespo J., Integration of Relational and Textual Biomedical Sources, Methods Inf Med 2009;48(1):76-83 [2] Tsung O. Cheng, Letters to Editor; in: Medical Abbreviations Journal of the Royal Society of Medicine, 97 (11), 2004: 556 [3] Semeck´ y J., Zv´ arov´ a J.(supervisor), Multimedia electronic health record in cardiology. Diploma thesis, Faculty of Mathematics and Physics of Charles University in Prague, 2001 (in Czech) [4] Smart Open Services for European Patients, D3.2.2 Final definition of functional service requirements – Patient Summary, www.epsos.eu (last access 14.7.2015)
65
66
ˇ Zivicov´ a V. et al. – Stroma of Head and Neck Squamous Cell Carcinoma
Stroma of Head and Neck Squamous Cell Carcinoma ˇ Veronika Zivicov´ a1,2 , Zdenˇ ek F´ık1,2 , Barbora Dvoˇr´ ankov´ a2 , Karel Smetana Jr.2 1 2
Institute of Anatomy, First Faculty of Medicine, Charles University, Prague, Czech Republic
Department of Otorhinolaryngology and HNS, First Faculty of Medicine, Charles University and Motol University Hospital,Czech Republic
ˇ Veronika Zivicov´ a Institute of Anatomy, First Faculty of Medicine, Charles University in Prague Address: U Nemocnice 3, 128 08 Prague 2 E–mail:
[email protected]
Aims of Research Among head and neck tumors there dominate squamous cell carcinomas (90%), arising from a mucous membrane of the upper respiratory tract and the upper part of digestive system, oropharynx, hypopharynx, and larynx – head and neck squamous cell carcinoma (HNSCC) [2, 7]. A supporting tissue of a malignant transformed epithelium is the tumor stroma, which consists of extracellular matrix, cancer associated fibroblasts (CAFs), myofibroblasts and other cell types (pericytes, smooth muscle cells, adipocytes, macrophages, and lymphocytes) [9]. The myofibroblast differs from the fibroblast in presence of contractile microfilaments. Reliable combination of signs, which is considered to identify the myofibroblast, are smooth muscle actin (SMA), P4H, vimentin, and absence of cytokeratins. Myofibroblasts influence negatively course of a chronic inflammation. In tumors, myofibroblast originates from CAFs, and induces disease progression [12]. This similary is included in Harald Dvorak’s theory about cancer like a wound, which does not heal [5]. A CAFs origin is still not clarified. It is considered the origin from local mesenchymal cells, from mesenchymal stem cells of bone marrow or from a cancer cell through epithelial-mesenchymal transition [4]. The other important component of the cancer stroma is fibronectin, sharing fibroblasts proliferation, immunologically competent cells chemotaxis, proteasis stimulation, etc. [11]. The presence of fibronectin in squamous cell carcinoma of oral cavity is associated with the presence of lymph node metastases and therefore it is also associated with a worse prognosis for patient [8, 3]. Tenascin is possibly detected in the extracellular matrix of HNSCC as well. Tenascin comprises a family of glycoproteins, from which tenascin-C is the best explored one. Tenascin-C plays a significant role in physiological process as well as in pathological condition. It was found increasingly not only in HNSCC, but in colorectal carcinoma, breast carcinoma, urinary bladder carcinoma,
lung cancer, and in glioma. One of its subunits (FN III) is able to bind fibronectin [10]. Tenascin together with glycosaminoglycans are researched nowadays in detail because of their role in cancer progression and as an possible target of anticancer therapy. [6, 1]. Presence of tenascin and its influence on tumor behaviour is evaluated in our laboratory. This project studies the tumor stroma under in vitro and in vivo conditions in correlation with clinical course. HNSCC samples were taken during surgery at ENT clinics based on informed consent from 54 patients. Frozen section technique was used and subsequently histological slides were prepared. Tenascin was detected in almost 80% cases. A healthy bucal mucous membrane was used as a control. According to our results, tenascin is restricted in the normal mucouse membrane to the bazal layer only. However, in squamous cell cancer, if positive, the whole stromal compartment was positive. In pursuit of patients dispenzarization, Kaplan-Meier survival analysis was performed. The analysis showed that tenascin positive patients survival is worse than in tenascin negative patients. However data are not statistically significant. Primary fibroblasts lines were gained from 4 tumors and 3 healthy controls. Primary fibroblasts were cultivated under standard tissue culture conditions and secondly stimulated by TGFb-1. In normal tissue culture conditions (for 7 days) normal fibroblasts do not produce tenascin. However stimulated cultures are tenascin positive. HNSSC cell lines do not respond to TGFb-1 stimulation in positive or negative manner. Very divergent examples of extracellular matrix production in HNSCC fibroblast cell lines are presented. This finding on HNSCC fibroblast lines can ilustrate heterogeneity of HNSCC seen in patients. Experiments confirm the important role of the tumor stroma in head and neck tumor biology. Its components like fibronectin or tenascin influence tumor growth and spread. Therefore it is necessary to know more about individual stromal components. These molecules have clinSemantic Interoperability in Biomedicine and Healthcare
ˇ Zivicov´ a V. et al. – Stroma of Head and Neck Squamous Cell Carcinoma
ical significance, because they could become appropriate MeSH: D005109 new targets of antitumor therapy. ICD10: not found
State of the Art
Tenascin
Head and neck squamous cell carcinomas (HNSCC) Definition: Hexameric extracellular matrix glycoprotein consist of tumor cells and the tumor stroma. Cancer transiently expressed in many developing organs and stroma is a tumor supporting tissue, which contains extraoften re-expressed in tumors. It is present in the cellular matrix, cancer associated fibroblasts (CAF) and central and peripheral nervous systems as well as in other cell types [9]. Even new therapies target specifically smooth muscle and tendons. against tumor cell, HNSCC often relapse and metastasize. The reason could be in the tumor stroma and its compo- Reference: not found nents like fibronectin or tenascin [11, 1]. SNOMED CT: not found
Application in Biomedicine and Healthcare
MeSH: D019063
Components of extracellular matrix play a significant role in the tumor biology. Matrix glycoproteins influence tumor growth and spread. Therefore these molecules could be used as new targets of antitumor therapy.
References
Acknowledgements This paper has been partially supported by the SVV2015-260158 project of Charles University in Prague.
ICD10: not found
[1] Afratis N, Gialeli C, Nikitovic D, Tsegenidis T, Karousou E, Theocharis AD, Pavao MS, Tzanakakis GN, Karamanos NK. Glycosaminoglycans: key players in cancer cell biology and treatment. FEBS J. 2012; 279(7):1177-97. [2] Argiris A, Karamouzis MV, Raben D, Ferris RL. Head and neck cancer. Lancet. 2008; 371(9625):1695-709.
Keywords
[3] de Bondt RB, Nelemans PJ, Hofman PA, Casselman JW, Kremer B, van Engelshoven JM, Beets-Tan RG. Detection of lymph node metastases in head and neck cancer: a meta-analysis comparing US, USgFNAC, CT and MR imaging. Eur J Radiol. 2007; 64(2):266-72.
Carcinoma
[4] De Wever O, Demetter P, Mareel M, Bracke M. Stromal myofibroblasts are drivers of invasive cancer growth. Int J Cancer. 2008; 123(10):2229-38.
Definition: Malignant epithelial neoplasm Reference: not found SNOMED CT: 68453008 MeSH: D002277 ICD10: not found
Extracellular matrix Definition: A meshwork-like substance found within the extracellular space and in association with the basement membrane of the cell surface. It promotes cellular proliferation and provides a supporting structure to which cells or cell lysates in culture dishes adhere. Reference: Kreis & Vale, Guidebook to the Extracellular Matrix and Adhesion Proteins, 1993, p93 SNOMED CT: not found
Semantic Interoperability in Biomedicine and Healthcare
[5] Dvorak HF. Tumors: wounds that do not heal. Similarities between tumor stroma generation and wound healing. N Engl J Med. 1986; 315(26):1650-9. [6] Guttery DS, Shaw JA, Lloyd K, Pringle JH, Walker RA. Expression of tenascin-C and its isoforms in the breast. Cancer Metastasis Rev. 2010;29(4):595-606. [7] Licitra L, Felip E, Group EGW. Squamous cell carcinoma of the head and neck: ESMO clinical recommendations for diagnosis, treatment and follow-up. Ann Oncol. 2009; 20 Suppl 4:121-2. [8] Lyons AJ, Bateman AC, Spedding A, Primrose JN, Mandel U. Oncofetal fibronectin and oral squamous cell carcinoma. Br J Oral Maxillofac Surg. 2001; 39(6):471-7. [9] Polyak K, Haviv I, Campbell IG. Co-evolution of tumor cells and their microenvironment. Trends Genet. 2009; 25(1):30-8. [10] Pas J, Wyszko E, Rolle K, Rychlewski L, Nowak S, Zukiel R, Barciszewski J. Analysis of structure and function of tenascin-C Int J Biochem Cell Biol. 2006;38(9):1594-602. [11] Ritzenthaler JD, Han S, Roman J. Stimulation of lung carcinoma cell growth by fibronectin-integrin signalling. Mol Biosyst. 2008; 4(12):1160-9. [12] Van Buerden HE, Von den Hoff JW, Torensma R, Maltha JC, Kuijpers-Jagtman AM. Myofibroblasts in palatal wound healing: prospects for the reduction of wound contraction after cleft palate repair. J Dens Res. 2005;84(10):871-80.
67
Poznámky:
Poznámky:
Poznámky:
Sémantická interoperabilita v biomedicíně a zdravotnictví
Editoři: Štěpán Svačina a Jana Zvárová Podpořeno projektem SVV-2015-260158
Fotografie na titulní straně: Karel Meister Grafický návrh: Anna Schlenker