Parametric inference for an extended illness-death model in the presence of censored data

Abstract

An extended illness-death model is a multistate model that offers an insightful approach to study the progression of successive events over a period of time. The anatomization of such a model includes inference on transition probabilities and transition hazard rate functions. In this paper, we consider an extended illness-death model with Weibull transition hazard and provide maximum likelihood and Bayes point and interval estimates of transition probabilities in the presence of randomly censored time-to-event data. Bayesian estimation is performed under non-informative and subjectively elicited priors. An extensive simulation study is conducted for numerical illustrations. Real data, concerning the study of hospital-acquired infection, are analysed. © 2025 Informa UK Limited, trading as Taylor & Francis Group.