多类型复发事件下的一类均值加速回归模型  

A Class of Accelerated Means Regression Models for Multiple Type Recurrent Event Data

作　　者：张启贤[1] 刘吉彩[2] 

机构地区：[1]三明学院信息工程学院,福建三明365004 [2]上海师范大学数理学院,上海200234

出　　处：《数学的实践与认识》2015年第5期179-188,共10页Mathematics in Practice and Theory

基　　金：国家青年自然科学基金(11401341);福建省教育厅高等学校教学质量工程项目(ZL0902/TZ(SJ));福建省教育厅科技项目(JA12301);三明学院教育教学改革项目(L1109/Q)

摘　　要：在复发事件的统计分析中,事件的平均发生个数可能比其强度函数或危险率函数更具可解释性.为了评价协变量对复发事件的影响,许多学者考虑了复发事件的边际比例均值模型.然而,在许多实际应用中,协变量对复发事件不仅具有均值比例效应,而且还可能会加速或减缓复发事件的发生,即协变量对复发事件均值过程具有时间尺度效应.在多类型复发事件数据框架下,考虑一类广泛的加速均值模型.利用估计方程方法,获得了该模型中未知参数的估计,并且建立了所给估计的渐近性质.进一步通过模拟研究证实所提方法的优良表现.In analysis of recurrent event, the mean number of recurrences may be more inter- pretable than the intensity function or the hazard function. Various mean models have been developed by many authors to assess the effects of covariates on the recurrent events. How- ever, inmany settings, the effects of covariates may not only alter the relative ratio between mean functions, but also accelerate or decelerate the recurrence of the related event. In this paper, we study a more general class of accelerated means models for multiple type recurrent event data to characterize the effect of covariates. The estimates of unknown parameters and baseline mean functions are obtained by a generalized estimating equation method and their asymptotic properties are established. A simulation study is conducted to illustrate the proposed procedure.

关 键 词：生存分析 多类型复发事件 加速均值回归模型 边际模型 广义估计方程 半参数推断 

分 类 号：O212.1[理学—概率论与数理统计]