Joint Modeling of Longitudinal Measurements and Multiple Failure Time Using Fully-specified Subdistribution Model: A Bayesian Perspective

Abstract

In biomedical studies, competing risks framework in which an individual fails due to multiple causes is frequently available. Joint modeling of longitudinal measurements and competing risks has become prominent, recently. In this paper, we proposed a joint model considering fully- specified subdistribution model introduced by Ge and Chen (2012) and longitudinal measurements. The proposed model links a linear mixed effect submodel to a fully-specified subdistribution submodel through a shared random effect. A Bayesian paradigm using MCMC is adopted to estimate the parameters. Performance of the proposed model is illustrated using a simulation study. In addition, this model is used to analyze the lung transplant dataset.