A Selective Review of Individualized Decision Making

With fast development in data-driven decision science, we have seen great advances in individualized decision making, including precision medicine, personalized marketing, and policy making. In particular, for precision medicine, due to the inherent heterogeneity among people, different patients may respond differently to a particular treatment. Compared to the one-size-fits-all strategy, it is more desirable to identify the optimal treatment among available options for every individual. Given data with covariates, treatment assignments and outcomes, a key goal is to find the optimal individualized decision rule that maps the individual characteristics or contextual information to the treatment assignment, such that the overall expected outcome is optimized. In this article, we provide a comprehensive review of the development on individualized decision making, where we particularly focus on the applications in precision medicine. Both single-stage and multi-stage problems are considered, with a selective review on several existing methods in the literature. We also discuss strengths and weaknesses of these methods, and offer some suggestions in handling real-world applications.