Learning Optimal Personalized Treatment Rules Under Risk Constraint

• Activity Number: 579
• Type: Invited
• Date/Time: Wednesday, August 12, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Yuanjia Wang* and Haoda Fu and Donglin Zeng
• Companies: Columbia University and Eli Lilly and Company and The University of North Carolina
• Keywords: Personalized Treatment Regimen ; Personalized Health Care ; Support Vector Machine ; Benefit risk analysis

Abstract

With novel technologies and multiple treatment options for the same disease, personalized medicine has become an important issue in the new era of medical product development. However, personalized clinical decision making is often complex since each treatment may exhibit distinct safety and efficacy profile for different patient populations. An optimal treatment for a patient that maximizes clinical benefit may also lead to greater concern of safety and adverse events. Thus, to guide individualized clinical decision making and deliver optimal tailored treatments, maximizing clinical benefit should be considered in conjunction of controlling for the risk. We propose a new outcome weighted learning approach to identify personalized optimal treatment strategy that maximizes clinical benefit under a constraint for the risk. Algorithms, simulations, and theoretical properties will be presented. We apply our approach to a randomized trial of type 2 diabetes to guide optimal administration of the first line insulin treatments based on individual patient characteristics while controlling for the number of hypoglycemia events.