A 2-step estimation procedure for semiparametric mixture cure models

TL;DR

This paper introduces a two-step estimation method for semiparametric mixture cure models that improves estimation accuracy in small samples by combining nonparametric presmoothing with parametric projection.

Contribution

It proposes a novel two-step procedure for estimating cure probabilities, enhancing performance over traditional maximum likelihood methods in small samples.

Findings

The new method outperforms existing estimators in simulations.

Theoretical properties of the estimator are established.

Application to melanoma data demonstrates practical utility.

Abstract

Cure models have been developed as an alternative modelling approach to conventional survival analysis in order to account for the presence of cured subjects that will never experience the event of interest. Mixture cure models, which model separately the cure probability and the survival of uncured subjects depending on a set of covariates, are particularly useful for distinguishing curative from life-prolonging effects. In practice, it is common to assume a parametric model for the cure probability and a semiparametric model for the survival of the susceptibles. Because of the latent cure status, maximum likelihood estimation is performed by means of the iterative EM algorithm. Here, we focus on the cure probabilities and propose a two-step procedure to improve upon the performance of the maximum likelihood estimator when the sample size is not large. The new method is based on the idea of presmoothing by first constructing a nonparametric estimator and then projecting it into the desired parametric class. We investigate the theoretical properties of the resulting estimator and show through an extensive simulation study for the logistic-Cox model that it outperforms the existing method. Practical use of the method is illustrated through two melanoma datasets.