Personalizing AI Interventions in Multiple Health Behavioral Change Settings

Samantha Marks, Michelle Chang, Eura Nofshin, Weiwei Pan, Finale Doshi-Velez

Keywords: reinforcement learning, RL, mobile health, mHealth, multiple health behavioral change, MHBC, personalization

Abstract: We introduce a novel reinforcement learning (RL) framework for personalizing AI interventions in multiple health behavior change (MHBC) settings. Our key contribution is a simple, interpretable model that captures empirically observed human behaviors. Using this model, we provide insight into how the AI will intervene, including when it has varying degrees of knowledge about the human model.

Submission Number: 79