Nonparametric cure rate estimation for multivariate failure times
Cure rate models play a critical role in survival analysis by identifying subpopulations no longer at risk of experiencing the event of interest, with applications in diverse fields such as oncology and clinical dentistry. When failure times are multivariate, as with multiple correlated events within subjects, current models fall short in addressing dependencies within subjects and estimating a shared cure rate across correlated outcomes. This paper develops a nonparametric framework for joint cure rate estimation in multivariate survival data, extending the Kaplan-Meier estimator to allow for intra-subject dependencies without parametric assumptions. We introduce two hypothesis tests, a two-sample test and a Wald-type test, to compare cure rates across margins in both bivariate and multivariate settings. Through simulation studies, we evaluate the estimator and tests under various dependency structures and sample sizes. The method's applicability is further illustrated using a dental restoration longevity study, demonstrating its capability to yield efficient survival estimates for multivariate data. The proposed methods are implemented in the R package multiCure.
Authors
- Jason P. Fine (ORCID: https://orcid.org/0000-0001-7208-7983)
- Zhenlong Xu (ORCID: https://orcid.org/0009-0006-4450-1693)
- Jun Yan (ORCID: https://orcid.org/0000-0003-4401-7296)
Institutions
- University of Connecticut (US)
- University of Pittsburgh (US)
Publication Details
- Journal
- Statistical Methods in Medical Research
- Published
- 2026-11-18
- DOI
- https://doi.org/10.1177/09622802261506991
- Primary Topic
- Statistical Distribution Estimation and Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00