鞅在分层比例风险模型中的应用

Application of martingale in stratified proportional hazards model

用极大似然估计的方法确定带有右删失数据的具有Cox强度的时间相依协变量分层比例风险模型的参数估计形式,并采用鞅方法得出该模型的参数估计是相合的且是渐近正态的.由此得出,在临床试验中该模型的参数采用极大似然法来估计是可行的.

This paper uses the maximum likelihood estimate method confirm the form of parametric estimators for the Cox regression model for censored survival data with time-dependent covariates which have a stratum proportional effect on the hazard function of the life-time distribution of an individual, and taking the martingale analysis method gained the estimators are consistent and asymptotic normal. This result shows that using maximum likelihood method to estimate the parameters of the model is feasible.