A1970
Title: A framework of Polya-Gamma Gibbs sampler for analyzing panel counts and terminal event time data
Authors: Chia-Hui Huang - National Chengchi University (Taiwan) [presenting]
Abstract: A joint analysis of panel counts and terminal event time models is developed under the Bayesian framework. The counts and event status are collected at discrete points, where the two types of events may exhibit either positive or negative dependence. The mean function of the counts and hazard rate of the terminal event are linked by frailty variables whose covariance matrix determines the underlying association. A Gaussian process is applied to model the logarithm of the baseline mean and hazard rate functions. Under the Bayesian framework, the Gibbs sampler with Polya-Gamma latent variables provides an efficient Markov chain Monte Carlo method to obtain posterior samples. The proposed method is applied to a clinical trial dataset consisting of panel counts and a terminal event.
