Bayesian Random Forest for Predictive Biomarkers Identification

• Activity Number: 144
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Wei Zheng* and Yuefeng Lu and Yang Zhao
• Companies: Sanofi and Sanofi and Sanofi
• Keywords: Predictive biomarkers ; Bayesian regression trees

Abstract

Predictive biomarker identification is an important topic in personalized medicine and clinical trials for nowadays targeted therapies. We present a Bayesian random forest method to model the biomarkers to assess the predictive impact on the clinical outcomes. Our method models the main effects and interactions separately, and provides predictive impact scores of the biomarkers for future enrichment study designs. The simulations show that our proposed method outperforms the conventional predictive biomarker identification methods by providing a higher power.