Characterization-based goodness-of-fit tests for survival data with a cure fraction

Abstract

We revisit a goodness-of-fit problem for randomly right-censored data with cured subjects, where the population consists of a non-susceptible group and those at risk of the event given sufficient follow-up. Modified U-statistics for such data are proposed, and their asymptotic properties are explored. This approach is illustrated through modifications of recent characterization-based goodness-of-fit tests for the exponential distribution. Their competitiveness is demonstrated through results from an extensive empirical study.