A0976
Title: A presmoothed estimator for the cure probability: An application to a cardiotoxicity dataset
Authors: Ana Lopez-Cheda - University of A Coruna (Spain) [presenting]
Samuel Saavedra - University of A Coruna (Spain)
M Amalia Jacome - Universidade da Coruna (Spain)
Abstract: Current cancer treatments have caused an increased ratio of cured patients or, at least, long-term survival. To accommodate the insusceptible proportion of subjects, a cure fraction can be explicitly incorporated into survival models, and as a consequence, cure models will arise. The goals in cure models are usually to estimate the cure rate and the probability of survival of the uncured patients up to a given point in time (latency). Although, in the literature, parametric and semiparametric models have been considered, nonparametric estimation methods for cure models have attracted much attention in the last few years. A presmoothed nonparametric estimator is proposed for the probability of cure in mixture cure models. Specifically, the methodology in a prior study is considered to improve the cure rate estimator of a subsequent study. The introduced nonparametric estimator is compared with existing approaches in a simulation study. Finally, the proposed methodology is applied to a study of cardiotoxicity in breast cancer patients.
