利用三次样条函数考察Cox模型比例风险假定

Assessing Proportional Hazards Assumption in Cox Regression Models Using Cubic Spline Functions.

目的 介绍一种检查Cox模型比例风险假定的假设检验方法。方法 利用时间的三次样条函数评价Cox比例风险回归模型中的时协变量交互作用项。结果 该法灵活有效 ,并且提供LHRF的点估计和区间估计。结论 三次样条回归作为一种检验方法 ,可与其他检验方法或图法结合使用 。

Objective To introduce a test for assessing proportional hazards assumption in Cox regression models.Methods We provide a method which investigates the time-by-covariates interactions in Cox proportional hazards regression models using cubic spline functions.Results The methods is flexible and effective.It also provides LHRF and it's confidence bands.Conclusion As a test,cubic spline regression can be used with other tests or graphical methods to investigate the proportional hazards assumption in Cox regression models.