LLM-driven coaching prompts grounded in the Transtheoretical Model of Change increase daily step count compared to generic prompts in adults at risk for cardiovascular disease.
This study will test whether personalized text-based coaching prompts generated by a large language model (LLM) increase daily step count more than generic prompts. The primary endpoint is change in daily step count over each intervention week in a randomized crossover trial with 15,000 adults across the US and UK. Recruitment began in March 2026 and is ongoing.
This article is automatically generated from the structured evidence profile behind the claim above. Scores reflect the quality and quantity of available research, not clinical advice.
LLM-driven coaching prompts grounded in the Transtheoretical Model of Change increase daily step count compared to generic prompts in adults at risk for cardiovascular disease. The current body of evidence comprises 1 study, including 1 randomized controlled trial. EvidenceHub rates the overall confidence at 44/100 (low).
The Claim
LLM-driven coaching prompts grounded in the Transtheoretical Model of Change increase daily step count compared to generic prompts in adults at risk for cardiovascular disease.
This conclusion is most relevant to: Adults (N=15,000) across the United States and United Kingdom, recruited via smartphone application.
What the Research Shows
The conclusion draws on 1 linked study. Highlights from the cited literature:
- ▸Design and rationale of the my heart counts cardiovascular health study: a large-scale, fully digital biobank, and randomized trial of large language model-driven coaching of physical activity. (American journal of preventive cardiology, 2026) —
How It Works
The proposed biological pathway:
- ▸LLM generates personalized text-based coaching prompts tailored to user's stage of change
- ▸Prompts are delivered via smartphone application daily or weekly
- ▸User receives prompts and is encouraged to increase physical activity (e.g., steps)
- ▸Result: Increased daily step count compared to generic prompts
Who Might Benefit
Evidence fit by population:
- ▸Adults (N=15,000) across the United States and United Kingdom, recruited via smartphone application
Recommended Dose
N/A
Limitations & Caveats
Important context when interpreting this evidence:
- ▸Results are not yet available; the study is ongoing with recruitment starting March 2026
- ▸Potential for selection bias due to smartphone-based recruitment and iOS-only initial release (Android planned for 2027)
- ▸Reliance on self-reported and passively collected mobile health data may introduce measurement error
Frequently Asked Questions
What is the primary outcome of the My Heart Counts randomized trial?▼
The primary outcome is the change in daily step count between the LLM-driven coaching prompts and generic text-based prompts, analyzed using mixed-effects models.
How many participants are being recruited for this study?▼
The target enrollment is 15,000 adults across the United States and United Kingdom.
What type of coaching is being tested in the embedded trial?▼
The trial tests personalized text-based coaching prompts generated by a large language model (LLM) grounded in the Transtheoretical Model of Change, compared to generic prompts.
When did recruitment for the My Heart Counts study begin?▼
Recruitment commenced in March 2026 and is ongoing.
References
- 1.Schmiedmayer P, Johnson A, Schuetz N, Kollmer L, Goldschmidt P, Delgado-SanMartin J, Zhang KW, Mantena SD, Tolas A, Montalvo S, Ramirez-Posada M, O'Sullivan JW, Oppezzo M, King AC, Rodriguez F, Ashley E, Lawrie A, Kim DS. “Design and rationale of the my heart counts cardiovascular health study: a large-scale, fully digital biobank, and randomized trial of large language model-driven coaching of physical activity..” American journal of preventive cardiology, 2026. PMID: 42395096 DOI: 10.1016/j.ajpc.2026.101565