|Title||Workshop on Personalized Medicine and Dynamic Treatment Regimes|
|Year of Publication||2012|
|Authors||Davidian, Marie, Michael R. Kosorok, Eric B. Laber, and Anastasios A. Tsiatis|
Personalized medicine is focused on making treatment decisions for an individual patient based on his/her clinical, genetic, and other characteristics. Traditional approaches to this goal seek to develop new treatments that are tailored to specific subgroups of patients with unique characteristics. An alternative objective is to determine the best treatment for each and every patient, not only those in a small subgroup, to the benefit of the entire patient population.
This workshop will take this point of view and introduce basic concepts and methods for discovery of individualized treatment regimes based on data. In the simplest case of a single treatment decision, a treatment regime is a rule that assigns treatment to patients based on their own patient-level characteristics, and the goal is to find the optimal regime, that leading to the greatest benefit to individual patients and the population if followed by all patients. In practice, treatment decisions may be made at multiple time points, as in the case of cancer treatment. Dynamic treatment regimes are thus series of sequential decision rules that dictate how to treat a patient over time, and the optimal regime is the set of rules that assigns treatment at each time point so as to maximize long term clinical outcome.
The workshop will first provide a broad overview of these ideas. The case of a single treatment decision, including definition of the optimal regime and statistical methods for estimating optimal regimes based on data, will then be discussed. Extension to the case of multiple decisions will follow, and statistical learning tools such as Q-learning and more recent developments for estimating optimal regimes from data will be introduced. Techniques for designing Phase II and Phase III clinical trials focused on discovery and verification of individualized treatment regimes will also be discussed.
The presentation will assume no prior exposure to the material.
Participants should have a good working knowledge of standard statistical methods, including regression analysis.