Application Of Competing Risk Longitudinal Data Modeling In Chronic Disease Health Management

Chronic disease has seriously threatened the health of Chinese people,and has become a major public health issue endangering the economic and social development of the country.Through comprehensive monitoring of individual or group patients,health management of chronic diseases can screen and assess the health status and risk factors of disease development.It can provide guidance and effect evaluation of chronic diseases with reasonable disease prediction model,which is of great significance to help improve patients’ treatment plan and strengthen self-management ability.In view of the frequent occurrence of competitive risk events in the development of chronic diseases,this article establishes a joint model for competitive risk events to help relevant researchers make a more reasonable explanation and explanation on how to analyze the longitudinal data under competitive risk,so as to better serve the health management of patients with chronic diseases.This article describes the correlation between time-to-event in the survival data of competing risk by using the cause-specific hazard model under competing risk as the survival sub-model firstly and analysis the longitudinal measurement process by using the linear mixed-effects model as the longitudinal sub-model subsequently.Then this article combines the two models by sharing random effect to construct the joint model of cause competing risk,and estimates the parameters of the model based on the maximum likelihood estimation method.In order to explore the stability of the model and identify the impact of competition risk on target outcome events,longitudinal simulation data samples with different competition proportions are generated to verify the model performance.Taking the standard error of parameter estimation and 95% confidence interval coverage rate as the evaluation criteria of point estimation and interval estimation,this article explores the differences of joint model of cause specific competing risk under different competing risk proportions,and compares the results with Cox proportional hazards model.The results indicate that the larger the proportion of competing risk,the larger the standard error of the regression coefficient estimation and the wider of the 95% confidence interval gradually.When the proportion of competitive risk is fixed,the joint model of cause-specific competing risk is better than the time varying covariate Cox model.Furthermore,taking the data of primary biliary cirrhosis(PBC)as an example,we explore the application of the model in real data.Firstly,the survival data of PBC were analyzed by using cause-specific hazard model and sub-distribution hazard model,and the difference of parameter estimation and model fitting effect between the two competing models were compared by using the parameter estimation standard error,cumulative risk function and ROC curve.Then the individual survival risk is predicted by means of the histogram.In addition,based on PBC longitudinal data,we construct a causespecific risk joint model and explore its statistical analysis performance under different survival outcomes and the differences with the cause-specific survival model.Results show that ignoring competing risk would lead to an overestimation of mortality risk in PBC patients;The classical competing risk model has a good prediction effect on the survival data of PBC,and the difference is not significant when the competing proportion is small;The parameter estimation of the joint model and the significance of the influencing factors under different outcomes may vary to some extent,so the existence of competitive risk cannot be ignored in the actual clinical analysis.Considering the existence of competing risk,a new competing risk joint model is constructed by combining the competing risk model with the longitudinal data model which deals with repeated measurement data.Combined with simulation test data and experiment data,this article discusses the practical application and software implementation of the joint model and expects to supply the analytical support for follow-up studies on the existence of competitive outcomes and provide some assistance for the health management of chronic diseases patients.