Random Power Function of Bent-cable Growth Models for Longitudinal Data: Application to AIDS Studies

Abstract

This paper presents a power function within a bent-cable model for identifying a transition period for the development of drug resistance to antiretroviral treatment (ART) in HIV/AIDS studies or other related fields. Such a transition period usually occurs between incoming and outgoing phases of trajectories of longitudinal data. The transition curve is modeled using a power function with unknown width, and we assess this clinically important feature of the longitudinal HIV/AIDS data using the bent-cable framework with random subject-specific parameters, including parameters of the transition curve. We develop a fully Bayesian approach for fitting a power function of bent-cable Tobit model to left-censored data and illustrate the proposed methods on real data from an AIDS clinical study.