Artificial intelligence applications in exercise and physical activity for public health: a scoping review

BackgroundPhysical activity and exercise are important components of public health strategies for preventing and managing chronic conditions and promoting health and quality of life. Advances in artificial intelligence (AI) have broadened its applications in this field, from personalized interventions and monitoring to risk prediction, professional decision support, and planning. This scoping review aimed to map the literature on AI-based exercise and physical activity applications from a public health perspective, characterize their functional roles, and examine the scope and maturity of the research.MethodsThis scoping review followed the PRISMA Extension for Scoping Reviews (PRISMA-ScR) and JBI methodological guidance. Web of Science, Scopus, and PubMed were searched for eligible studies published between 2015 and 2025. Two Reviewers independently conducted title and abstract screening and full-text assessment, resolving disagreements by consensus. Of 612 records identified, 601 remained after duplicate removal and 24 studies met the eligibility criteria. The included studies were characterized by publication year, country, study design, AI technology and function, and stage of evidence development. Key findings were synthesized across the identified functional categories.ResultsAmong the included studies, publication activity was highest in 2024, while China accounted for the largest number of studies. Five functional categories were identified: Intervention Delivery or Adaptation, Monitoring or Assessment, Prediction or Risk Stratification, Professional Decision Support, and System and Environmental Planning. Research maturity differed across these functions. Technical development and validation studies often reported promising performance, whereas intervention findings were less consistent and did not always show additional benefit over comparison conditions. Many studies remained at the model-development, technical-validation, feasibility, or pilot stageConclusionAI is being used across several functions in exercise and physical activity, although evidence of consistent clinical or public health benefit remains limited. Current findings support its complementary use in personalization, monitoring, prediction, and professional decision support; the available findings do not justify autonomous clinical decision-making. Further controlled studies, external validation, longer follow-up, and real-world evaluation are needed to establish whether technical capabilities translate into sustained physical activity and meaningful health outcomes. Implementation should also account for professional oversight, data protection, and the needs of diverse populations.