Multivariate Survival Analysis via Model Selection

Recently, several frailty models have been developed and widely used for analyzing multivariate survival data. Generally, selecting a suitable model among a set of candidate models is very important in data analysis. In this paper we propose a model selection criterion for semi-parametric frailty models. The proposed method is based on the conditional likelihood, a component of hierarchical likelihood and is an extension of conditional Akaike information(Vaida and Blanchard, 2005) for mixed linear models to the frailty models. The proposed method is illustrated using a well-known real data set. Here we show how to select and analyze an appropriate model