B0195
Title: Optimizing the dynamic personalized health care decision rules when clinical restrictions exist
Authors: Lu Wang - University of Michigan (United States) [presenting]
Abstract: Recent advances and statistical developments are presented for evaluating dynamic treatment regimes (DTR), which allow the treatment to be dynamically tailored according to evolving subject-level data. Identification of an optimal DTR is a key component for precision medicine and personalized health care. A tree-based doubly robust reinforcement learning (T-RL) method is presented, which builds an unsupervised decision tree that maintains the nature of batch-mode reinforcement learning, and then a new stochastic-tree search method called ST-RL for evaluating optimal DTRs, which contributes to the existing literature in its non-greedy policy search and demonstrates outstanding performances even with a large number of covariates. In addition, a common challenge is considered with practical "restrictions" on the treatment sequences: (i) one or more treatment sequences that were offered to individuals when the data were collected are no longer considered viable in practice; (ii) specific treatment sequences are no longer available; or (iii) the scientific focus of the analysis concerns a specific type of treatment sequences (e.g., "stepped-up" treatments). To address this challenge, a restricted tree-based reinforcement learning (RT-RL) method is developed. The method is illustrated using an observational dataset to estimate a two-stage stepped-up DTR for guiding the level of care placement for adolescents with substance use disorder.
