Multistate frailty model for interval-censored data under Bayesian framework

Pawan Lamsal, Johannes Kepler University Linz

Co-authors: Helga Wagner, Johannes Kepler University Linz; Magdalena Muszynska-Spielauer; Austrian Academy of Sciences

Abstract: Multistate frailty models for interval censored data are used to model the risk of transition between observed states over time by incorporating unobserved individual characteristics called frailties. Such models can be fitted by estimating the parameters in a frequentist approach by maximizing the marginal likelihood function of the observed data  integrating out the frailty, which  however is not trivial. In this study, we use a  Bayesian method by implementing a Markov chain Monte Carlo (MCMC) sampling scheme to sample from the posterior distribution of the model parameters. Since observed living states are interval censored, we apply a piecewise constant hazard approximation to fit the model to data from the Survey of health, aging and retirement in Europe (SHARE) for Germany, present and discuss the findings.