A Class of Semiparametric Mixture Cure Survival Models with Prevalent Survival Data

• Activity Number: 42
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Yu-Jen Cheng*+
• Companies: National Tsing Hua University
• Keywords: mixture cure models ; survival ; prevalent sampling

Abstract

A class of semiparametric mixture cure survival models with prevalent survival data is considered in this article, where semiparametric transformation models are used to estimate the survival function of uncured subjects, and the cure rate is estimated from a logistic regression. In contrast to the existing works, we focus on survival data collected from a prevalent cohort study which is a well-known biased sampling. Without correcting such sampling bias, the naive estimation will lead to a severe bias for both parameters in the semiparametric transformation models and logistic regression. A pseudo partial likelihood implementing with EM or multiple imputation procedure is proposed. The proposed approaches are examined through stimulation studies and applied to a real data.