Title: Bayesian spatial homogeneity pursuit for survival data
Authors: Lijiang Geng - University of Connecticut (United States) [presenting]
Guanyu Hu - University of Missouri Columbia (United States)
Abstract: A new Bayesian spatial homogeneity pursuit method is proposed for survival data under the Cox proportional hazards model, to detect spatially clustered patterns in the associations between hazard rates and covariates. Specially, regression coefficients and baseline hazard are assumed to have spatial homogeneity over space. To capture the homogeneity in regression coefficients and baseline hazards, we develop a geographically weighted Chinese restaurant process prior for them. An efficient Markov chain Monte Carlo (MCMC) algorithm is designed to estimate the clustered coefficients and baseline hazards and their uncertainty measures simultaneously. Extensive simulations are conducted to evaluate the empirical performance of the proposed models. Finally, we illustrate the performance of the model with a real data analysis of respiration cancer in Louisiana.