A0243
Title: Bayesian semi-parametric mixed hidden Markov models
Authors: Kai Kang - Sun Yat-sen University (China) [presenting]
Abstract: A semi-parametric mixed hidden Markov model is developed to analyze longitudinal data. The proposed model comprises a parametric transition model for examining how potential predictors influence the probability of transition from one state to another and a non-parametric conditional model for revealing the functional effects of explanatory variables on outcomes of interest. Unlike conventional regression that focuses only on the observation process, the proposed model simultaneously investigates the observation process and the underlying transition process. Two correlated random effects, the one is in the conditional model and the other is in the transition model, are considered to describe the possible dependency within and/or between the two stochastic processes. We propose a Bayesian approach that combines Bayesian P-splines and MCMC methods to conduct the statistical analysis. The empirical performance of the proposed methodology is evaluated via simulation studies. An application to a real-life example is presented.
