Online Program Home
My Program

Abstract Details

Activity Number: 632 - Statistical Issues Specific the Therapeutic Areas-4
Type: Contributed
Date/Time: Thursday, August 2, 2018 : 8:30 AM to 10:20 AM
Sponsor: Biopharmaceutical Section
Abstract #328887 Presentation
Title: Variational Inference for Proportional Hazards Model with Power Prior in Oncology Studies
Author(s): Bo Jin* and Yue Chang
Companies: Boston Biomedical Inc. and Boston Biomedical Inc
Keywords: Hazard Ratio; Variational approximation; MCMC approximation; Bayesian; Power prior

In this paper we consider the problem of analyzing survival data from a randomized and standard therapy-controlled early phase oncology study. We take Bayesian approaches with power prior to incorporate historical data of the standard therapy. We assume that the hazard functions are proportional over time and specify the baseline hazard function by a piece-wise constant hazard model. The talk's contribution is to develop a variational approximation (VA) approach to estimation and prior power selection for this model. We conduct simulation experiments to compare the VA approach to Monte Carlo Markov Chain (MCMC) approximation to the posterior and illustrate the application of VA in model selection. VA produces point estimates with comparable accuracy with MCMC at faster speed and provides a quick method for model selection.

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

Back to the full JSM 2018 program