A Bayesian Approach for Tailored Therapy Strategy: Hidden Markov Mixture ITP Model for Predicting Patient Response to Therapy

• Activity Number: 233
• Type: Contributed
• Date/Time: Monday, August 3, 2009 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Wei Chen*+ and Haoda Fu
• Companies: University of Michigan and Eli Lilly and Company
• Keywords: Tailored therapy ; Two-compnent Prediction Model ; Hidden Markov Model ; Patient response

Abstract

Variability in patient response to drugs is a significant therapeutic issue. It can lead to sub-optimal drug reactions in sub-populations of patients. Therefore, finding tailored therapy to fit individual profiles has become great interest in recent pharmaceutical research. We propose a novel Bayesian approach for predicting patient response to therapy, which will help better discriminate responders and nonresponders to the drug. A key part of this approach is, instead of predicting the patient response to therapy once at the end of the trial, we can predict the response at each trial time point dynamically by imposing a Hidden Markov structure on the mixture Integrated Two-component Prediction model (Fu, Manner 2008). In this way, we offer additional information for differentiating patient respondent groups to the drug, which will help offer best-fit therapies to patients ultimately.