Chatbot as an intervention on physical activity: A systematic review and meta-analysis with meta-regression

Background: Chatbots offer scalable, accessible, and personalized support for physical activity (PA). This review synthesized evidence on their efficacy and examined potential moderators. Methods: Eleven databases were searched for English-language published through January 27, 2026. Eligibility was assessed using the PICOS framework. Data extraction was guided by the research questions, mERA checklist, and PRISMA 2020. Risk of bias was evaluated using Cochrane Risk of Bias 2 tool and the NIH Quality Assessment Tool. Meta-analyses reported Hedges' g. Publication bias was evaluated using funnel plots, Egger's regression, and trim-and-fill analyses. Potential moderators were examined through subgroup analyses and meta-regression. Results: A total of 12,481 records were identified, of which 43 studies were included in the systematic review and 27 in the meta-analysis. Most studies included in the meta-analysis had some risk-of-bias concerns (16/27). Chatbot interventions produced a significant small-to-moderate post-intervention effect on PA (g = 0.31, 95% CI [0.19, 0.44]; n = 24), whereas the follow-up effect was small and non-significant (g = 0.11, 95% CI [-0.01, 0.22]; n = 4). Effects varied across PA outcomes (g = 0.08-0.44), and several subgroup differences helped explain the significant heterogeneity. For daily steps, significant subgroup differences were observed by clinical status, while intervention duration (QM = 6.09, p = 0.048) and clinical status (QM = 5.69, p = 0.017) were significant moderators. Conclusions: Chatbot interventions showed small-to-moderate post-intervention effects but limited follow-up effects. Few significant moderators were identified, underscoring the need for more rigorous RCTs.

Authors

Institutions

Publication Details

Journal
medRxiv
Published
2026-11-16
DOI
https://doi.org/10.64898/2026.11.13.26358176
Primary Topic
Mobile Health and mHealth Applications
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Chatbot as an intervention on physical activity: A systematic review and meta-analysis with meta-regression

Dongmin Song, Yang Bai, Lingyi Fu
2 links
medRxiv
Mobile Health and mHealth Applications
preprint

Chatbot as an intervention on physical activity: A systematic review and meta-analysis with meta-regression

Dongmin Song, Yang Bai, Lingyi Fu
preprint en

Abstract

Background: Chatbots offer scalable, accessible, and personalized support for physical activity (PA). This review synthesized evidence on their efficacy and examined potential moderators. Methods: Eleven databases were searched for English-language published through January 27, 2026. Eligibility was assessed using the PICOS framework. Data extraction was guided by the research questions, mERA checklist, and PRISMA 2020. Risk of bias was evaluated using Cochrane Risk of Bias 2 tool and the NIH Quality Assessment Tool. Meta-analyses reported Hedges' g. Publication bias was evaluated using funnel plots, Egger's regression, and trim-and-fill analyses. Potential moderators were examined through subgroup analyses and meta-regression. Results: A total of 12,481 records were identified, of which 43 studies were included in the systematic review and 27 in the meta-analysis. Most studies included in the meta-analysis had some risk-of-bias concerns (16/27). Chatbot interventions produced a significant small-to-moderate post-intervention effect on PA (g = 0.31, 95% CI [0.19, 0.44]; n = 24), whereas the follow-up effect was small and non-significant (g = 0.11, 95% CI [-0.01, 0.22]; n = 4). Effects varied across PA outcomes (g = 0.08-0.44), and several subgroup differences helped explain the significant heterogeneity. For daily steps, significant subgroup differences were observed by clinical status, while intervention duration (QM = 6.09, p = 0.048) and clinical status (QM = 5.69, p = 0.017) were significant moderators. Conclusions: Chatbot interventions showed small-to-moderate post-intervention effects but limited follow-up effects. Few significant moderators were identified, underscoring the need for more rigorous RCTs.

medRxiv
University of Utah (US)
Good health and well-being
Mobile Health and mHealth Applications
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.

Laika access coming soon