Fine-tuning LLMs in behavioral psychology for scalable health coaching

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 on the Transtheoretical Model of Change. MHC-Coach generates messages tailored to an individual's psychology (their "stage of change"), providing personalized support to foster long-term physical activity behavior change. To evaluate MHC-Coach's efficacy, 632 participants compared human-expert and MHC-Coach interventions encouraging physical activity. Among messages matched to an individual's stage of change, 68.0% (<i>N</i> = 430) preferred MHC-Coach-generated messages (<i>P</i> < 0.001). Blinded behavioral science experts (<i>N</i> = 2) rated MHC-Coach messages higher than human-expert messages for perceived effectiveness (4.4 vs. 2.8) and Transtheoretical Model alignment (4.1 vs. 3.5) on a 5-point Likert scale. This work demonstrates how language models can operationalize behavioral science frameworks for personalized health coaching, showing the potential for promoting long-term physical activity and reducing cardiovascular disease risk at scale.

Authors

Institutions

Publication Details

Journal
npj Cardiovascular Health
Published
2025-09-22
DOI
https://doi.org/10.1038/s44325-025-00083-5
Citations
3
Primary Topic
Behavioral Health and Interventions
Type
article
Field-Weighted Citation Impact
4.63

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Fine-tuning LLMs in behavioral psychology for scalable health coaching

Anders Johnson, Daniel Seung Kim, ­Abby C. King, Narayan Schütz et al.
3 citations
npj Cardiovascular Health
Behavioral Health and Interventions
4.63
article

Fine-tuning LLMs in behavioral psychology for scalable health coaching

Anders Johnson, Daniel Seung Kim, ­Abby C. King, Narayan Schütz, Mariana Ramírez‐Posada, Fátima Rodríguez, Alexander Tolas, Marily Oppezzo, Paul Schmiedmayer, C. Mikael Mattsson, Ritu Doijad, Euan Angus Ashley, Eleni Linos, Allan Lawrie, Sriya Mantena
article en
3 citations

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 on the Transtheoretical Model of Change. MHC-Coach generates messages tailored to an individual's psychology (their "stage of change"), providing personalized support to foster long-term physical activity behavior change. To evaluate MHC-Coach's efficacy, 632 participants compared human-expert and MHC-Coach interventions encouraging physical activity. Among messages matched to an individual's stage of change, 68.0% (<i>N</i> = 430) preferred MHC-Coach-generated messages (<i>P</i> < 0.001). Blinded behavioral science experts (<i>N</i> = 2) rated MHC-Coach messages higher than human-expert messages for perceived effectiveness (4.4 vs. 2.8) and Transtheoretical Model alignment (4.1 vs. 3.5) on a 5-point Likert scale. This work demonstrates how language models can operationalize behavioral science frameworks for personalized health coaching, showing the potential for promoting long-term physical activity and reducing cardiovascular disease risk at scale.

npj Cardiovascular HealthVol. 2(1)
University of Washington (US), Cardiovascular Institute of the South (US), Stanford Health Care (US), Stanford Medicine (US), Imperial College London (GB), Stanford University (US)
American Diabetes Association, American Heart Association, University of Washington, Imperial College London, Canada Excellence Research Chairs, Government of Canada, National Institutes of Health, School of Medicine, Stanford University, Stanford Bio-X, National Heart, Lung, and Blood Institute
Openalex Percentile: Top 5%
Behavioral Health and Interventions
4.63
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.