The nonparametric location-scale mixture cure model

TL;DR

This paper introduces a fully nonparametric approach for modeling mixture cure models with censored data, providing a root-n consistent estimator of the error distribution that simplifies bandwidth selection.

Contribution

It develops a novel nonparametric estimator for the error distribution in mixture cure models that is asymptotically independent of bandwidth choice, reducing computational complexity.

Findings

Estimator is root-n consistent under certain bandwidth conditions

Simulation shows good finite sample performance

Application to breast cancer data illustrates practical utility

Abstract

We propose completely nonparametric methodology to investigate location-scale modelling of two-component mixture cure models, where the responses of interest are only indirectly observable due to the presence of censoring and the presence of so-called long-term survivors that are always censored. We use covariate-localized nonparametric estimators, which depend on a bandwidth sequence, to propose an estimator of the error distribution function that has not been considered before in the literature. When this bandwidth belongs to a certain range of undersmoothing bandwidths, the asymptotic distribution of the proposed estimator of the error distribution function does not depend on this bandwidth, and this estimator is shown to be root-n consistent. This suggests that a computationally costly bandwidth selection procedure is unnecessary to obtain an effective estimator of the error distribution, and that a simpler rule-of-thumb approach can be used instead. A simulation study investigates the finite sample properties of our approach, and the methodology is illustrated using data obtained to study the behavior of distant metastasis in lymph-node-negative breast cancer patients.