Title: Nonparametric survival estimation with missing not at random censoring indicators
Authors: Olivier Goudet - University of Angers (France)
Mikael Escobar-Bach - University of Angers (France) [presenting]
Abstract: In the presence of right-censored data with random covariates, the conditional Kaplan-Meier estimator (also referred to as the Beran estimator) consistently estimates the conditional survival function. However, it relies on the knowledge of each individual censoring status, which might be missing in practice. We thus show a study for the Beran estimator when the censoring indicators are not clearly specified, and next, propose a new method for the conditional survival function estimation with missing not at random (MNAR) censoring indicators. Along with the theoretical results, we illustrate how the estimators work for small samples by means of a simulation study and show their practical applicability with the analysis of synthetic data.