Fine-tuning Large Language Models in Behavioral Psychology for Scalable Physical Activity Coaching DOI Creative Commons

Sriya Mantena,

Anders Johnson, Marily Oppezzo

et al.

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 21, 2025

ABSTRACT Personalized, smartphone-based coaching improves physical activity but relies on static, human-crafted messages. We introduce My Heart Counts (MHC)-Coach, a large language model fine-tuned the Transtheoretical Model of Change. MHC-Coach generates messages tailored to an individual’s psychology (their “stage change”), providing personalized support foster long-term behavior change. To evaluate MHC-Coach’s efficacy, 632 participants compared human-expert and text-based interventions encouraging activity. Among matched stage change, 68.0% (N=430) preferred MHC-Coach-generated ( P < 0.001). Blinded behavioral science experts (N=2) rated higher than for perceived effectiveness (4.4 vs. 2.8) alignment (4.1 3.5) 5-point Likert scale. This work demonstrates how models can operationalize frameworks health coaching, promoting potentially reducing cardiovascular disease risk at

Language: Английский

Fine-tuning Large Language Models in Behavioral Psychology for Scalable Physical Activity Coaching DOI Creative Commons

Sriya Mantena,

Anders Johnson, Marily Oppezzo

et al.

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 21, 2025

ABSTRACT Personalized, smartphone-based coaching improves physical activity but relies on static, human-crafted messages. We introduce My Heart Counts (MHC)-Coach, a large language model fine-tuned the Transtheoretical Model of Change. MHC-Coach generates messages tailored to an individual’s psychology (their “stage change”), providing personalized support foster long-term behavior change. To evaluate MHC-Coach’s efficacy, 632 participants compared human-expert and text-based interventions encouraging activity. Among matched stage change, 68.0% (N=430) preferred MHC-Coach-generated ( P < 0.001). Blinded behavioral science experts (N=2) rated higher than for perceived effectiveness (4.4 vs. 2.8) alignment (4.1 3.5) 5-point Likert scale. This work demonstrates how models can operationalize frameworks health coaching, promoting potentially reducing cardiovascular disease risk at

Language: Английский

Citations

0