A0793
Title: Contextual bandits for micro-randomized trials in mobile health
Authors: Bibhas Chakraborty - Duke-NUS Medical School, National University of Singapore (Singapore) [presenting]
Abstract: Mobile health (mHealth) interventions (e.g., motivational text messages or nudges to promote healthy behaviors) are becoming increasingly common in tandem with advances in mobile and wearable sensor technologies. An innovative trial design arising in mHealth is discussed, namely, the micro-randomized trial (MRT) that involves sequential, within-person randomization over many instances. The basic MRT design can be further improved to make it adaptive, thereby enabling it to learn from accumulated data as the trial progresses. This is appealing from an ethical perspective since adaptive learning tends to make better interventions available to the trial participants. Adaptive learning in such trial designs is often operationalized via contextual bandit algorithms. Specifically, the role of a particular algorithm called Thompson sampling in designing adaptive MRTs is discussed. Theoretical as well as simulation results will be shown to validate the proposed approach. mHealth clinical trials will be discussed as case studies.
