Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 287 - Current and Future Challenges in Analyzing Composite Endpoints
Type: Invited
Date/Time: Wednesday, August 5, 2020 : 10:00 AM to 11:50 AM
Sponsor: Lifetime Data Science Section
Abstract #309514
Title: Joint Bayesian Nonparametric Models for Survival Times and Medical Costs
Author(s): Jason Roy* and Arman Oganisian and Nandita Mitra
Companies: Rutgers University and University of Pennsylvania and University of Pennsylvania
Keywords: causal inference; Bayesian nonparametrics; joint models

Comparative cost effectiveness analysis requires estimates for the causal effect of treatment on both cost and effectiveness (e.g., survival). However, joint modeling of cost and survival are challenging due to censoring of total costs, censoring of survival times, zero-inflated outcomes, skewed outcomes, and confounding. Here we propose a joint model based on a gamma process prior for survival and an enriched Dirichlet process prior for costs. Marginal estimates of causal effects are obtained after integrating over the marginal distribution of confounders using Bayesian bootstrap. This approach enables inference on the joint posterior of cost and effectiveness. The methodology is illustrated using data on cost effectiveness of cancer therapies.

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

Back to the full JSM 2020 program