Personalized wellness at scale: A framework for AI-driven feedback in mHealth

This paper examines the potential and challenges of artificial intelligence (AI) agents in mHealth (mobile health) wellness technology. Specifically, we explore how AI can enhance personalization and feedback, two design features users consistently value but that are difficult to scale. We discuss how modern AI agents can extend existing feedback mechanisms and encourage user engagement. We also examine safety and ethical concerns raised by AI-generated wellness feedback. Drawing on prior mHealth and AI personalization research, we propose a framework for responsibly embedding AI-driven feedback into wellness applications. As a proof-of-concept instantiation, we developed Wally, a prototype AI-powered wellness assistant. A preliminary assessment using three simulated agents found that Wally performed satisfactorily for a low-risk user but produced clinically inappropriate responses for a higher-risk profile. We conclude by identifying design refinements and future research needed to assess the ethical, practical, and empirical challenges of AI-enabled wellness support.