(magistrsko delo)
Jasmina Rebernak (Avtor), Uroš Potočnik (Mentor), Ivan Ferkolj (Komentor)

Povzetek

Naredili smo celovito študijo s podatki iz članka (Koder et al., 2015) in doktorata (Koder, Silvo 2015). Namen naše raziskave je bila uporaba različnih statističnih pristopov (t-test neodvisnih vzorcev, anova, test dveh neodvisnih vzorcev, test k neodvisnih vzorcev, korelacije, linearna in logistična regresija) in z njimi odkriti povezave med biokemijskimi, kliničnimi in genetskimi podatki z odzivom na zdravljenje bolnikov s CB z adalimumabom. Uporabili smo programsko orodje SPSS in test normalne porazdelitve ShapiroWilk za izvedbo statističnih analiz. Odziv smo merili z vprašalnikom IBDQ in biokemijskim parametrom C-reaktivni protein (CRP). Za vsakega pacienta (N = 97) smo beležili 24 biokemijskih, 34 kliničnih in 34 genetskih podatkov. S kliničnim odzivom IBDQ se je dvakrat ponovilo pet polimorfizmov pri posameznemu genu. Trikrat so se ponovili trije biokemijski in šest kliničnih parametrov. Najbolj signifikantne povezave smo zaznali s PTGER4, CASP9, CCNY, sakroileitisom in alkalno fosfatazo. Z biološkim odzivom CRP se je dvakrat ponovilo devet povezav s polimorfizmi pri posameznem genu. Trikrat sta se ponovila dva biokemijska in eden klinični parameter. Za najbolj signifikantne so se izkazale povezave z ATG16L1, IL27, NR1|2 in c11orf30 ter albuminom. Z uporabo linearne in logistične (binarne) regresije smo izdelali napovedne modele za IBDQ in CRP. Z linearno regresijo smo dobili model z najvišjim R2 (0,691) za IBDQ v tridesetem tednu (natančnost: 69,32 %). Medtem ko smo najboljšo natančnost dobili za IBDQ v dvanajstem (76,00 %, R2 = 0,619) in dvajsetem (75,68 %, R2 = 0,567) tednu. Modeli za CRP so imeli R2 nad 0,980, vendar je njihova natančnost pod 61 %. Pri logistični regresiji je največjo natančnost imel model za IBDQ v četrtem tednu (R2 = 0,912; natančnost: 81,48 %). Natančnost modelov za odziv v dvanajstem (R2 = 0,482), dvajsetem (R2 = 0,465) in tridesetem (R2 = 0,653) tednu je znašala 71,43 %, 68,75 % in 64,10 %. Natančnost modelov za odziv CRP je znašala nad 70 %, vendar je to posledica manjkajočih podatkov in s tem manj veljavnih vzorcev. Ob koncu študije smo izdelali tabelo, s katero smo predstavili izbiro statističnega testa glede na biokemijske, klinične in genetske podatke.

Ključne besede

crohnova bolezen;adalimumab;statistične metode;Mann-Whitney;Fisherʼs Exact test;Chi square test;test neodvisnih vzorcev;linearna regresija;logistična regresija;napovedni model;

Podatki

Jezik: Slovenski jezik
Leto izida:
Tipologija: 2.09 - Magistrsko delo
Organizacija: UM FZV - Fakulteta za zdravstvene vede
Založnik: [J. Rebernak]
UDK: 616.3:575(043.2)
COBISS: 2156964 Povezava se bo odprla v novem oknu
Št. ogledov: 1539
Št. prenosov: 271
Ocena: 0 (0 glasov)
Metapodatki: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Ostali podatki

Sekundarni jezik: Angleški jezik
Sekundarni naslov: STATISTICAL APPROACHES FOR DETECTING ASSOCIATIONS AMONG BIOCHEMICAL, GENETIC AND CLINICAL DATA WITH TREATMENT RESPONSE OF CROHN´S DISEASE PATIENTS WITH ADALIMUMAB
Sekundarni povzetek: We used article (Koder et al., 2015) and doctorate (Koder Silvo, 2015) data to carry out a comprehensive study. Purpose of our study the use of different statistical methods (independent samples t-test, one-way anova, 2 independent samples test, k independent samples test, correlations, linear and logistic (binary) regression) to detect associations among biochemical, clinical and genetic data with treatment response of CD patients with adalimumab. Statistical program package SPSS and ShapiroWilk’s tests of normality were used to carry out statistical analyses. Response was measured using IBDQ questionnaire and biochemical marker C-reactive protein (CRP). We collected 24 biochemical, 34 clinical and 34 genetic data for each patient (N = 97). Clinical response IBDQ was associated with 5 genetic polymorphisms that were repeated at least two times. Minimum three significant repetitions were discovered with three biochemical and six clinical parameters. PTGER4, CASP9, CCNY, sakroiliitis and alkaline phosphatase were among most significant associations. Biological response CRP was associated with 9 genetic polymorphisms, three biochemical and one clinical parameter. Among most significant associations were ATG16L1, IL27, NR1|2, c110rf and albumin. Linear and logistic (binary) regressions were used to construct prediction models for IBDQ and CRP. Highest R2 (0,691) in linear regression was associated with model for IBDQ in week 30 (accuracy: 69,32 %). Highest accuracy was calculated for IBDQ model in week twelve (76,00 %; R2 = 0,619) and twenty (75,68 %; R2 = 0,567). Models for CRP had R2 above 0,980 while accuracy was below 61 %. Highest accuracy in logistic regression was calculated for response IBDQ in week 4 (R2 = 0,912; accuracy: 81,48 %). Model accuracy for response IBDQ in week twelve (R2 = 0,482), twenty (R2 = 0,465) and thirty (R2 = 0,653) were 71,43 %, 68,75 % and 64,10 %. Models for response CRP had less valid samples due to missing data, therefore the accuracy was above 70 %. At the end of our study we assembled a table demonstrating the use of statistical tests based on biochemical, clinical and genetic data.
Sekundarne ključne besede: Crohnʼs disease;adalimumab;statistical methods;Mann-Whitney;Fisherʼs Exact test;Independent samples T-test;Chi square;linear regression;logistic regression;prediction model;
URN: URN:SI:UM:
Vrsta dela (COBISS): Magistrsko delo/naloga
Komentar na gradivo: Univ. v Mariboru, Fak. za zdravstvene vede
Strani: XIV, 129 str.
ID: 8776568