The Joint Model Of Longitudinal And Semi-competing Risks Data

In clinical trials,longitudinal data and semi-competing risks data are often collected from the same patient.The combined analysis of these data will improve the efficiency of statistical inference.Therefore,we proposes a new joint model of longitudinal data and semi-competing risks data in this paper.In this model,the longitudinal sub-model is assumed to be a linear mixed effect model,the survival sub-model is assumed to be a illness-death model,and the two submodels are correlated by the shared random effects.Because of the random effect,marginal likelihood involves integration and has no closed set form,so we give an adjust profile likelihood by means of the Laplace approximation.The parameters are estimated by the quasi-Newton method,and the simulation results are given.Simulation results are provided to validate performance of the proposed joint model.Compared with other models,the joint model has superior performance in terms of bias and standard error.We also applied this model to the study of patients with metastatic colorectal cancer to provide some insights for the prognosis management of patients.