ABSTRACT

Treatment Regime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485 15.4 Challenge 2: Identifying Patient Subgroups with Maximal

Benefit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488 15.5 Challenge 3: Accounting for Multivariate Outcomes . . . . . . . . . . . 490 15.6 Challenge 4: Quantifying and Communicating Uncertainty

Associated with Treatment Recommendations . . . . . . . . . . . . . . . . 492 15.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495

ABSTRACT A treatment regime formalizes personalized treatment selection as a function that maps available patient information to a recommended treatment. Treatment regimes, by recommending if, what, when, and to whom treatment should be applied, have the potential for better patient outcomes as well as lower cost and patient burden. Thus, the goals of treatment regimes are closely aligned with those of comparative effectiveness research. However, many of the practical considerations central to comparative effectiveness research such as clinical impact, cost-benefit analyses, and subgroup identification have received little attention in the treatment regime literature. In this chapter, we review regression-based estimators of optimal treatment regimes. We use this class of estimators to illustrate four key challenges associated with evaluating the clinical and practical benefits of personalized treatment regimes; we offer some preliminary solutions and discuss several pressing open problems.