Sampling-based inference for the generalized time-dependent logistic hazard model

Abstract

Lifetimes which satisfy a non-proportional hazard model may arise in several areas, such as, Medicine, Biometrics, Criminology and Industrial Reliability. For these data it is reasonable to presume that the hazard function is time-dependent, thereby accommodating crossing hazards. Such dependency can be modelled directly by introducing a time-dependent term in the model for the hazard function. Accordingly, in this paper we utilize a generalized time-dependent logistic (GTDL) hazard model which can accommodate non-proportional hazards data. A sampling-based inference procedure based on Markov chain Monte Carlo Methods is developed and the methodology is used to investigate survival from advanced lung cancer in a well known dataset.