Statistical Inference Of Competing Risk Data Based On Semi-parametric Model And Its Application

In recent years,the study of competitive risk data has received widespread attention from scholars.This article studies the statistical inference and application of competitive risk data based on a semi-parametric model,mainly including the following two aspects.The first content of this article is the joint hypothesis testing problem of competitive risk data.In clinical trials,sub-distribution risk and all-cause risk are important indicators of therapeutic efficacy and have high research value.However,there have been no reports on the joint testing of two sub-distribution risks and the joint testing of all-cause risks and sub-distribution risks in competitive risk data based on semi-parametric models.Therefore,this article proposes two joint testing methods for sub-distribution risks and a joint testing method for all-cause risks and sub-distribution risks based on right censored competition risk data and interval censored competition risk data.In this paper,the semi-parametric cumulative incidence functions of two failure types are given,the nonparametric test programs of two joint test problems are established,and the asymptotic properties of the test statistics are given.The results show that the proposed method performs well through extensive simulation studies.Finally,the proposed method is applied to a set of real data to illustrate.The second part of this study is to study the outcome of radical surgery with or without adjuvant radiotherapy for primary squamous cell carcinoma of the breast based on a semi-parametric model.Due to the fact that patients with primary squamous cell carcinoma of the breast may have different outcomes after treatment,and these outcomes compete with each other.Previous studies have regarded patients who die from other causes as missing,which can bias statistical results.Therefore,competitive risk models need to be used for statistical analysis.This article views death from primary squamous cell carcinoma of the breast as an outcome event,and death from other causes as a competitive event.The competitive risk model was used for univariate analysis and multivariate analysis of etiological specific survival rate to analyze prognostic factors and increase the accuracy of the study results.Kaplan-Meier method was used to calculate the etiologic specific survival rate and overall survival rate,and survival curves were drawn.The Lasso Cox regression model was constructed to analyze the independent risk factors of overall survival rate,so as to evaluate the effects of treatment methods on etiologic specific survival rate and overall survival rate.The results showed that lumpectomy combined with adjuvant radiotherapy significantly improved the survival rate and could be used as a treatment option for primary squamous cell carcinoma of the breast.