Bayesian modeling of clustered competing risks survival times with spatial random effects

This is the authors' abstract. We don't add key points for this paper.

n some studies, survival data are arranged spatially such as geographical regions. Incorporating spatial associationin these data not only can increase the accuracy and efficiency of the parameter estimation, but it also investigatesthe spatial patterns of survivorship. In this paper, we considered a Bayesian hierarchical survival model in thesetting of competing risks for the spatially clustered HIV/AIDS data. In this model, a Weibull Parametric distributionwith the spatial random effects in the form of county-failure type-level was used. A multivariate intrinsic conditionalautoregressive (MCAR) distribution was employed to model the areal spatial random effects. Comparison amongcompeting models was performed by the deviance information criterion and log pseudo-marginal likelihood. Weillustrated the gains of our model through the simulation studies and application to the HIV/AIDS data.