Stanford researchers discovered something counterintuitive: people respond better to AI-generated motivational messages about exercise than to advice crafted by human health experts. The finding comes from My Heart Counts, a decade-long cardiovascular research initiative that has engaged over 100,000 study participants and just launched an upgraded version with a new AI coaching system.

What Makes AI Messages More Effective Than Expert Advice?

The difference lies not in the content itself, but in how the message is tailored to each individual. Most fitness apps simply track your activity and send generic reminders. My Heart Counts takes a fundamentally different approach by using behavioral psychology principles to understand where each person stands in their readiness to change.

The AI coaching system draws on the Transtheoretical Model of Change, a framework developed in the 1980s that describes how people move through distinct stages when adopting new behaviors. Someone who has never considered exercising needs a completely different message than someone who already jogs three times a week but risks slipping back into old habits. The AI identifies where each user sits on that spectrum and generates tailored messages accordingly.

"People can know how important exercise is and also feel they don't have time to exercise. We are hoping this app, grounded in decades of behavioral research, can support people in their journey to health, whether they are just beginning or already on their way," said Euan Ashley, MB ChB, DPhil, one of the dedicated clinician researchers on My Heart Counts' core team.

Euan Ashley, MB ChB, DPhil, Clinician Researcher at Stanford University School of Medicine

Heart disease remains the leading cause of death worldwide, a distinction it has held for decades despite extraordinary advances in medicine. The frustrating reality is that one of the most effective interventions, regular physical activity, is also one of the hardest to sustain. My Heart Counts was designed to study that gap and help people cross it.

How Does the AI Coaching System Actually Work?

• Behavioral Staging: The system identifies whether a user is in early stages of considering exercise, actively preparing to change, or already maintaining an exercise routine, then generates messages matched to that stage.
• Personalized Inputs: Messages are generated using structured inputs including a person's stage of change, basic contextual factors like age and physical limitations, preferred language, and preferred delivery time.
• Focused Design: The system is not a chatbot and does not engage in open-ended conversations or accept free-text input, design choices that keep the AI focused and its outputs predictable and grounded in behavioral science.
• Holistic Health Tracking: Beyond step counts, the app's Heart Health dashboard integrates activity data, sleep patterns, and clinical biomarkers from Apple Health, helping users understand factors like sleep and mental health that play a role in heart disease prevention.

"The AI isn't making medical decisions. It's generating short, encouraging activity messages using a structure grounded in behavioral science," explained Anders Johnson, Project Manager at the Department of Medicine.

Anders Johnson, Project Manager at Department of Medicine

The app launched in 2015 in partnership with Apple and captured data from over 40,000 participants in just two weeks. A decade later, the platform has engaged over 100,000 study participants and has just upgraded its interface with the new AI coaching study.

Why Should Healthcare Systems Care About This Finding?

The research has broader implications for how healthcare organizations think about patient engagement and digital health tools. If AI-generated messages consistently outperform human-crafted advice, it suggests that personalization and behavioral science matter more than the perceived authority or expertise of the message source. This challenges the assumption that patients always prefer human interaction in healthcare.

The finding also arrives at a critical moment in healthcare innovation. The global personalized medicine market is projected to reach USD 1.4 trillion by 2035, growing from USD 654.46 billion in 2025, with a compound annual growth rate of 7.88 percent. This expansion reflects a broader shift from reactive, one-size-fits-all approaches to proactive, AI-driven precision care.

The integration of AI with multi-omics data, which includes genomics, proteomics, and metabolomics, is revolutionizing healthcare by moving from generic treatments to hyper-personalized patient care. By synthesizing extensive datasets from wearable devices, electronic health records, and molecular profiling, these AI-driven platforms enable predictive disease trajectories and accelerate drug discovery, particularly in oncology and rare diseases.

My Heart Counts demonstrates that this personalization principle extends beyond diagnosis and treatment selection to the fundamental challenge of behavior change. The app's success suggests that the future of healthcare may depend less on having the smartest experts and more on having systems that understand individual psychology and deliver the right message at the right moment.