Bootstrap Variance Estimate For Marginal AFT Model

In this paper,the semi-parametric model of accelerated failure time model,which has no specified error distribution,is studied.For right-censored data,the process of the accelerated failure time model based on the least square method is introduced.The proposed parameter estimation method takes the initial consistent estimator Gehan estimation as the starting value,and the iterative solution obtained by iteration of the Buckley-James estimation equation.The estimator is consistent and asymptotically normal.The performance of the new parameter estimation method is evaluated by the simulation study.When dealing with data with small sample size,we increase the size of sample data through Bootstrap method to calculate the parameter estimator variance,compare the superiority of estimates of the variance.Heart transplantation data and multiple myeloma data were used for case analysis to estimate the parameters and standard deviation of the model and to analyze the significance of the estimated results.