Keywords
Bayesian inference; HIV/AIDS; piecewise model; time to event; skew-distribution
Abstract
This paper presents a new methodology for jointly identifying bent cable phasic patterns and mixture of progressors and non-progressors of human immunodeficiency virus infection and acquired immune deficiency syndrome (HIV/AIDS) patients based on longitudinal and time-to-event data. Using the longitudinal data, the bentcable model gives an estimate of a gradual transition period for the development of drug resistance to Antiretroviral (ARV) drug for treating HIV patients. In addition to finding such an estimate (phasic pattern identification), a two-part modeling is carried out to incorporate a relatively large percentage of left-censored data in the framework of joint analysis of time to event and longitudinal data. Even though there are some methods for separately analyzing time to event and longitudinal data, those methods may not be appropriate when time to event is dependent on the longitudinal outcome. A better approach is to extend a bent-cable To bit model that jointly incorporates patients who are potentially progressors to AIDS from those patients who do not, phasic changes of trajectories of viral load, and the association between the time to a decline of CD4/CD8 ratio and rates of change in viral load. The proposed methods are illustrated using real data from an AIDS clinical study.
Citation
Dagne GA. Bayesian Bent-Cable Tobit Models for Longitudinal and Survival Data: Application to AIDS Studies. SM J Biometrics Biostat. 2018; 3(1): 1025.