Paper No. 07-27
GO Roberts and LM Sangalli
Latent Diffusion Models for Event History Analysis
Abstract: We consider Bayesian hierarchical models for event history analysis, where the event times are modeled through an underlying diffusion process, which determines the hazard rate. We show how these models can be efficiently treated by means of Markov chain Monte Carlo techniques.
Keywords: Diffusion processes, survival analysis, parametrization of hierarchical models.