竞争风险数据和协变量相依权重下可加可乘的子分布风险率模型  

An Additive-Multiplicative Subdistribution Hazard Model with Covariates-Dependent Weight for Competing Risks Data

作　　者：李婉星 龙永红[1] 薛清水 Wan Xing;LI Yong;Hong LONG;Qing Shui XUE(School of Information, Renmin University of China, Beijing 100872, P. R. China;Ministry of Finance of the People＇s Republic of China, Beijing 100820, P. R. China)

机构地区：[1]中国人民大学信息学院,北京100872 [2]国家财政部,北京100820

出　　处：《数学学报（中文版）》2018年第3期353-374,共22页Acta Mathematica Sinica：Chinese Series

摘　　要：本文在竞争风险数据下提出一种灵活的含变系数的可加可乘的子分布风险率模型．通过对删失时间的风险函数建立Cox比例风险模型，得到调整后的与协变量相依的权重，在新权重下建立估计方程来估计模型参数，并获得了估计的大样本性质，同时提出了模型中协变量的时变效应的检验方法．通过数值模拟验证了所提方法的有限样本性质，结果表明所提方法可以大大降低估计偏差．最后，分析了一组淋巴滤泡细胞的竞争风险数据集来展示所提方法的实际应用效果．We propose a flexible additive-multiplicative hazard model allowing the additive time-varying effects of covariate for the subdistribution in a competing risks circumstance. For inference on the model parameters, weighted estimating equation approaches under an covariates-dependent adjusted weight by fitting the Cox propor-tional hazard model for the censoring distribution are established. In addition, large number properties and a goodness-of-fit test procedure are presented. The finite sample behavior of the proposed estimators is evaluated through simulation studies, estima- tors from the proposed method perform satisfactorily on reduction of the bias, and an application to a competing risks data set from a follicular cell lymphoma study is illustrated.

关 键 词：竞争风险 变系数可加可乘风险率模型 累积发生函数 估计方程 逆删失概率 加权 

分 类 号：O212.1[理学—概率论与数理统计]