JMIR Publications

ABSTRACT

Background:

Artificial intelligence (AI) chatbots are technologies that facilitate human‒computer interaction through communication in a natural language format. By increasing cost-effectiveness, interaction, autonomy, personalization, and support, mHealth interventions can benefit health behavior change and make it more natural and intuitive.

Objective:

This study aimed to provide a comprehensive and up-to-date overview of the development (i.e., theories, techniques, and features) and use (i.e., roles, user experience, feasibility, efficacy, and effectiveness) of AI chatbots for health behavior change.

Methods:

In accordance with the PRISMA-ScR framework, relevant studies published before March 2024 were identified from nine bibliographic databases (i.e., PubMed, CINAHL, Medline, EMBASE, Web of Science, Scopus, APA PsycINFO, IEEE Xplore, and ACM Digital Library). Two stages (i.e., title and abstract screening followed by full-text screening) were conducted to screen the eligibility of the articles via Covidence software. Finally, we extracted the data via Microsoft Excel software and employed a narrative approach and content analysis to synthesize the reported results.

Results:

Our systematic search initially identified 10,508 publications, 43 of which met our inclusion criteria. The analysis revealed that physical activity was mostly targeted. The popular theories and techniques are cognitive behavioral therapy (CBT) and non-code platforms (e.g., Google Dialogflow and IBM Watson) integrated with social messaging platforms (e.g., Facebook Messenger). The AI chatbot dialogs were in a natural language format characterized by comprehensiveness, flexibility, variety, adaptability, empathy, support, and nonjudgment. It primarily served as coaches and self-monitoring and personalized support tools. Overall, the evaluation of these interventions revealed preliminary mixed findings for the user experience (i.e., acceptability, system utility, and system usability). In terms of efficacy, the AI chatbots effectively increased physical activity, improved dietary habits, reduced stress levels, promoted smoking cessation, and had a minor effect on weight management. However, significant evidence gaps remain, particularly concerning engagement, feasibility, efficacy for sleep improvement and alcohol use reduction, and effectiveness for health behavior change.

Conclusions:

The exploratory synthesis revealed mixed findings regarding certain user experience outcomes, alongside acceptable efficacy for certain health behaviors. Future research should employ methodologically rigorous designs, including randomized controlled trials and meta-analyses, to establish definitive conclusions about intervention efficacy and implementation potential. Clinical Trial: https://osf.io/pw7g9/