Semiparametric Bayesian Modeling of Multiple Changepoints in Longitudinal Data
by
Kaushik Ghosh
University of Nevada Las Vegas
Coauthors: Pulak Ghosh, Ram C. Tiwari

In this talk, we will present a new robust multiple-changepoint model for longitudinal data in the presence of informative dropouts. The model uses a Dirichlet process prior on the distribution of changepoints in individual trajectories. The inherent clustering property of Dirichlet processes leads to a natural clustering, and hence, sharing of information among subjects with similar trajectories, leading to improved estimates. A fully Bayesian approach is developed for model fitting and prediction. The proposed method is illustrated using data from a clinical trial.

Date received: March 13, 2008

Copyright © 2008 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # cawu-95.