A0775
Title: Accelerated failure time modelling via nonparametric mixtures
Authors: Byungtae Seo - Sungkyunkwan University (Korea, South) [presenting]
Sangwook Kang - Yonsei University (Korea, South)
Abstract: An accelerated failure time (AFT) model assuming a log-linear relationship between failure time and a set of covariates can be either parametric or semiparametric depending on the distributional assumption for the error term. Both classes of AFT models have been popular in the analysis of censored failure time data. The semiparametric AFT model is more flexible and robust to departures from the distributional assumption than its parametric counterpart. The semiparametric AFT model, however, is subject to producing biased results for estimating any quantities involving an intercept. Meanwhile, parametric AFT models can be severely impaired by misspecifications. We propose a new type of AFT model using a nonparametric Gaussian scale mixture distribution. The proposed method can provide a consistent and robust estimator for the structural parameters in AFT models. The finite sample properties of the proposed estimators will also be presented via an extensive simulation study and a read data set.
