Efficient Semiparametric Bayesian Estimation of Multivariate Discrete Proportional Hazards Model with Random Effects

Abstract

We incorporate a random effect into a multivariate discrete proportional hazards model and propose an efficient semiparametric Bayesian estimation method. By introducing a prior process for the parameters of baseline hazards, we consider a nonparametric estimation of baseline hazards function. Using a state space representation, we derive a dynamic modeling of baseline hazards function and propose an efficient block sampler for Markov chain Monte Carlo method. A numerical example using kidney patients data is given.

Keywords:

• Bayesian analysis
• Discrete survival time
• Markov chain Monte Carlo
• Multivariate proportional hazards
• Random effects
• State space model

Mathematics Subject Classification:

• Primary 62N02
• Secondary 62F15

Acknowledgment

The computational results are generated using Ox 4.04 (Doornik, 2002).