Accelerated failure time for Weibull distribution based partly interval censored data

Abstract

In this project, the performance of maximum likelihood estimators of the parameters of Accelerated Failure Time (AFT) regression model based on Weibull distribution with simple imputations methods under Partly-Interval Censored (PIC) data is studied and compared with semiparametric Cox model. From a real data set, the results indicate that the AFT with Weibull distribution is comparable with Cox model under PIC breast cancer data. This result suggests that the parameters of the model are stable and the treatments are significant for breast cancer patients. Hence, maximum likelihood estimation is an appropriate method for estimating the parameters of our Weibull AFT based on simple imputation method under PIC data. In the simulation study, using the AIC and LRT with their p-values, the results show that our model is fit well and flexible under PIC data especially for exact observation. This finding has led us to deduce the fact that the AFT with Weibull distribution can be useful for modeling PIC data