Seminar
Pulak Ghosh
Joint Modeling of Longitudinal Data and Informative Dropout in the Presence of Multiple Changepoints with an Application to HIV-AIDS
In longitudinal studies of patients with the Human Immunodeficiency Virus (HIV), objectives of interest often include modeling of individual-level trajectories of HIV Ribonucleic Acid (RNA) as a function of time. Such models can be used to predict the effects of different treatment regimens or classify subjects with similar trajectories into subgroups. This, in turn, helps in determining the optimal treatment combination(s), which is an important part of drug development. Empirical evidence, however, suggests that individual trajectories often possess multiple points of rapid change, which may vary from subject to subject. The modeling of individual viral RNA levels in the Presence of such changepoints becomes difficult, since usual methods become unsuitable. In this talk, we develop a new robust multiple-changepoint model which satisfactorily addresses the above issues. The proposed method uses a joint model to incorporate information from the longitudinal data as well as from informative dropouts, which are common in such studies. A Dirichlet process prior is used to model the distribution of the slopes associated with the changepoints of individual trajectories. The inherent clustering property of Dirichlet process leads to a natural clustering, and thus, sharing of information among subjects with similar trajectories. A fully Bayesian approach is used for model fitting and prediction and is implemented using the Gibbs sampler. The proposed method is illustrated using data from the ACTG 398 clinical trial. The results suggest that the proposed method is a significant improvement over the currently available methods.
