Leveraging User Feedback to Inform Just-in-time Adaptive Intervention (JITAI) Delivery in mHealth: An Exploratory Study

Just-in-time adaptive interventions (JITAIs) have become increasingly popular in mobile health (mHealth) technologies to promote health behavior changes. Despite their potential, one of the key challenges in applying JITAI in practice is the limited information these systems have about individual users. Personalized messaging depends on accurate and complete data about users’ behavior, context, and preferences. Inspired by interactive reinforcement learning and the demonstrated superiority of combining human input with algorithmic processes, our study explores collecting additional user feedback on received interventions to enrich data and improve algorithm training. We conducted a two-week Wizard of Oz study with 18 participants randomly assigned to three groups, each experiencing different feedback interaction designs. We logged their interactions, collected daily survey responses, and conducted exit interviews. Our findings provide insights into (1) user attitudes towards providing feedback and users’ feedback-providing behaviors and motivation; (2) users’ reasoning processes and logic behind providing feedback; and (3) design implications that support effective feedback-provision experiences. Our study contributes to the human-computer interaction (HCI) community by looking into the innovative idea of leveraging user feedback as a richer data source, offering the potential to enhance the personalization and effectiveness of mHealth interventions.