Two Joint Models Of Longitudinal And Time-to Event Data With Model Comparing Research Based On A Data Set Of Liver Cirrhosis Clinical Trial

Joint Model of longitudinal and time-to event data based on Cox model and linear mixed effect model has been widely used in a data set analysis of clinical trial,and some studies have pointed out that when the proportional risk assumption can not be well satisfied,modeling with Cox model will lead to deviations that can not be ignored.But in the practice of clinical trial data analysis,it is difficult to verify the hypothesis of the model,so it is an interesting and practical issue to compare the two kinds of joint models based on actual data.This paper based on the data of clinical trials of liver cirrhosis carried out by several hospitals in Copenhagen,a joint model based on Cox model and linear mixed effect model,and a joint model based on accelerated failure model and linear mixed effect model are constructed and compared,respectively.We have created a Joint model using the classical maximum likelihood approach for estimating parameter and using the EM approach for solving the complex integral problem then using the AIC and BIC for selecting variables,then using the mean square error for the accuracy of the two models in clinical data analysis compared.The results of this paper are as follows:The mean square error of the two models is very small,the AIC and BIC values are close,and the complexity of the models is similar in data analysis.Joint Model of longitudinal and time-to event data based on Cox model and linear mixed effect model takes less time to calculate parameters.So accelerated failure model can be a good substitute for Cox model.The purpose of this paper aims to provide a reference for clinical researchers choose survival models.All calculations of this paper have been completed by the statistical software R.