| The semiparametric accelerated failure time(AFT)model is not as widely used as Cox proportional hazard model due to computational difficulties.Recent developments in least squares estimation and induced smoothing estimating equations provide effective tools for censored data processing,which make the AFT models widely used.For marginal AFT models,we use the extended GEE method to censored data.Its advantage is that the consistency of the regression coefficient estimator is robust to misspecification of working covariance,and the estimation efficiency is higher when the working covariance structure is closer to the truth.The marginal error distributions and regression coefficients of each margin are allowed to be unique or partially shared across the margins as needed.The initial estimator is rank-based estimator with Gehan’ weights,but it is easier to calculate based on the induced smoothing method.The resulting estimator is consistent and asymptotically normal,and with variance estimated through a multiplier resampling method.In a large-scale simulation study,especially when the within-cluster dependence was strong,we considered the performance of the estimator obtained from binary continuous data and binary discrete data respectively.It has been proved that the estimator was up to three times as efficient as the estimator that ignores the within-cluster dependence.The methods were applied to the diabetic retinopathy data to detect the influences of single and interactive variables on the patient’s health. |
