Canal
Easy Estimation of Dynamic Treatment Regimes: A Primer on Personalized Medicine (303732)
*Michael Patrick Wallace, University of Waterloo
Keywords: causal inference, personalization, personalized medicine, decision making
Personalized medicine is a new (and quickly evolving) area of health research, focused on tailoring treatment decisions to patient-level characteristics. Dynamic treatment regimes (DTRs) formalize this process in a statistical framework, providing an analytical approach to finding the best treatment decisions across a population. Though their use has primarily been limited to health, DTR methodology may be applied to any situation where sequences of decisions are made to optimize some outcome, giving rise to numerous opportunities for their use beyond health-aligned fields. Despite this potential they have received limited use in practice, in part because of methodology that is often perceived (or presented) as complex, and a literature that has largely focused on medical contexts. This session will provide an introduction to the principles and processes of personalized medicine in general, and DTRs in particular, with no prior knowledge of the topic required. We will explain both the 'Why?' and 'How?' of this important topic, and demonstrate application through estimation methods that are not only easy to understand and use, but whose results can be readily communicated.