CMStatistics 2020: Start Registration
View Submission - CMStatistics
Title: Accelerated failure time vs Cox proportional hazards mixture cure models Authors:  Motahareh Parsa - KU Leuven (Belgium) [presenting]
Ingrid Van Keilegom - KU Leuven (Belgium)
Abstract: A cure model is a useful model for analyzing failure time data in which some subjects experience the event of interest (the uncured subjects), and others don't (the cured subjects). We focus on mixture cure models, which rely on a model for the cure probability and a model for the survival function of the uncured subjects, conditional on a set of covariates. For the latter model, one often uses a Cox proportional hazards model. Despite the many advantages of this model, like its easy interpretation and the availability of software, the model suffers from some important drawbacks, like the cure threshold is the same for all values of the covariates. This might be unrealistic in situations, where covariates contain important information about the cure proportion and the event time of subjects under study. An alternative model is the accelerated failure time (AFT) mixture cure model. The cure threshold in this model depends on the covariates and leads, therefore to a more realistic and better fit of the data in many cases. We show that the AFT and the Cox model both fit the data well in the regions of sufficient follow-up, but differ drastically outside that region.