带有辅助信息的多元失效时间数据的估计标准部分似然方法

Estimated standard partial likelihood method for multivariate failure time data with auxiliary information

在许多实际研究中,由于预算限制,主协变量值只能对某一个有效集进行准确测量,但同时对应此主协变量的辅助信息则对全部个体均可以观测.利用这些辅助协变量的信息有助于提高统计研究的效率.本文在基于共同基准危险率的边际模型框架下,我们提出了一些统计推断方法来分析多元失效时间数据.对于回归参数,我们提出标准的估计部分似然方程来估计它,同时也给出了累积基准危险率函数的Breslow型估计.得到的估计可以证明是相合的和渐近正态的.利用模拟分析结果来表明了提出的方法在有限样本下的可行性.

In some practical studies, the main covariate is only precisely measured in a validation set due to financial limitations, while as an auxiliary information for the main covariate are collected for the full cohort. Taking use of these auxiliary information will increase the study efficiency. In this paper, based on the framework of marginal proportional hazards model, we develop some inference procedures to analyze such multivariate failure time data. We propose an estimated standard partial likelihood equation for the regression parameters and a Breslow-type estimator for cumulative hazard function. The resultant estimators are shown to be consistent and asymptotically normally distributed. Simulation studies are conducted to evaluate the performance of the proposed estimators.