Joint Modeling Of Longitudinal Survival Data Of Liver Cirrhosis

In clinical research,joint modeling of longitudinal and survival data has been widely used,since it reduces bias in parameter estimation and increases the efficiency of statistical inference.We use Cox proportion hazards model for survival outcomes,a linear mixed model for continuous longitudinal outcomes,where these process are linked through shared parameters.In this paper,the standard joint model is extended,where the linear model is extended to the generalized linear model by using nonlinear function.Moreover,when the longitudinal results are continuous and polytomous variables,the cumulative logit regression is introduced to deal with the ordinal outcomes.The Cox proportional hazard model is combined with the generalized linear mixed model,and the maximum joint likelihood function are used for parameter estimation and statistical inference.We apply our method to PBC data analysis.By calculating the area under ROC curve(AUC)and the prediction deviation,we have done a comparative analysis between the standard joint model and the extended joint model.The results shows that the joint model for dealing with continuous and polytomous variables has better fitting goodness.