Abstract: Current-status data occur in many areas, and the analysis of such data attracted much attention. In this study, we consider a regression analysis of current-status data in the presence of informative censoring, for which most existing methods either apply only to limited situations or are computationally unstable. Here, we propose a new sieve maximum likelihood estimation procedure under the class of semiparametric generalized odds rate frailty models. The proposed method uses the latent variable to describe the informative censoring or relationship between the failure time of interest and the censoring time. We develop a novel expectation-maximization algorithm for determining the proposed estimators, and establish their asymptotic consistency and normality. The results of a simulation study show that the proposed method performs well in practical situations. In addition, we demonstrate the proposed method by applying it to a set of real data arising from a tumorigenicity experiment.

Key words and phrases: EM algorithm, generalized odds rate frailty models, informative censoring, sieve approach.