生存资料回归模型分析--Cox比例风险假设的拟合优度法

本文目的是介绍一些检验比例风险假设的方法。图示法是通过绘图然后由人工进行判断是否符合比例风险假设,因而具有一定的主观性。在图示法的基础上,本文介绍了从客观角度检验比例风险假设的一些常用方法,主要包括两类:一类是基于残差的检验;另一类则是构建协变量与时间的交互项并对其进行检验的方法。首先阐述了上述方法的原理,然后基于SAS软件并通过一个实例介绍上述方法的实现。

Analysis of regression model of survival data--the test of Cox’s proportional hazards assumption based on the goodness-of-fit

The main purpose of this study was to introduce some methods for the test of the proportional hazards assumption. The graphical methods of testing the assumption were subjective because they were plotted and then manually judged whether they met the proportional hazards assumption. Based on the graphical methods, this article introduced some common methods, based on the Goodness-of-fit, for testing the assumption of proportional hazards from an objective perspective, mainly including two types, one was based on the residual test, the other was to construct the interaction terms of covariates by time. First of all, the principle of the above method was explained, and then the implementation of the method in SAS software was introduced through an example.