Online Program Home
  My Program

Abstract Details

Activity Number: 503 - SPAAC Poster Competition
Type: Topic Contributed
Date/Time: Wednesday, August 2, 2017 : 10:30 AM to 12:20 PM
Sponsor: Scientific and Public Affairs Advisory Committee
Abstract #324799
Title: Bayesian Subgroup Finding by Stochastic Optimization
Author(s): Arinjita Bhattacharyya* and LURDES INOUE INOUE and Peter Mueller and Riten Mitra
Companies: University of Louisville and University of Washington and UT Austin and UNIVERSITY OF LOUISVILLE
Keywords: MCMC ; Bayesian ; utility ; inhomogeneous ; Weibull ; interactions

An important goal in clinical studies is to find subgroups of a population that respond differentially to treatment. In the context of clinical trials, efficient subgroup search would even lead to targeted therapy designs, thus furthering the goal of personalized medicine. We present a utility-based fully Bayesian approach for subgroup selection that a) finds the 'optimal' subgroup as well as b) an optimal family of subgroups. The utility function for both these aims is designed to satisfy some key heuristics and automatically incorporate issues of multiplicity appearing in (b). The standard MCMC is used for model inference in a), and subgroup inference is assessed by posterior expected utility. For (b), an optimization scheme based on inhomogeneous Markov Chains is applied. This stochastic method was primarily developed to search complex spaces and its application to subgroups is novel. This general approach to subgroup finding can be easily adapted to non-clinical settings and sociological data.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

Copyright © American Statistical Association