Premium
Censored survival models for genetic epidemiology: A gibbs sampling approach
Author(s) -
Gauderman W. James,
Thomas Duncan C.
Publication year - 1994
Publication title -
genetic epidemiology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.301
H-Index - 98
eISSN - 1098-2272
pISSN - 0741-0395
DOI - 10.1002/gepi.1370110207
Subject(s) - gibbs sampling , statistics , genetic epidemiology , sampling (signal processing) , econometrics , biology , epidemiology , mathematics , computer science , medicine , bayesian probability , filter (signal processing) , computer vision
Methods are proposed for the analysis of diseases with variable age at onset. The Cox proportional hazards model, widely used for epidemiologic analysis, is modified to include both measured (environmental) covariates and latent (genetic) variables, as well as their interactions. A Monte Carlo technique known as Gibbs sampling is utilized to generate observations from the posterior distributions of all model parameters. A correction to account for single ascertainment of pedigrees is also described. Simulation studies show that parameter estimation is nearly unbiased for a wide variety of models, and that moderate gene‐environment interaction effects can be detected. © 1994 Wiley‐Liss, Inc.