Maintenance Notice

Due to necessary scheduled maintenance, the JMIR Publications website will be unavailable from Wednesday, July 01, 2020 at 8:00 PM to 10:00 PM EST. We apologize in advance for any inconvenience this may cause you.

Who will be affected?

Previously submitted to: Journal of Medical Internet Research (no longer under consideration since Sep 15, 2026)

Date Submitted: Jun 25, 2026
Open Peer Review Period: Jun 25, 2026 - Aug 20, 2026
(closed for review but you can still tweet)

AI-Based Exercise Prescription for Chronic Disease Management: Umbrella Review of Systematic Reviews and Meta-Analyses

  • Yi-Ting Wang; 
  • Timon Chengyi Liu

Background:

AI-based exercise prescription is a fast-growing approach to chronic disease management, but the breadth and quality of the supporting evidence remain incompletely characterized.

Objective:

This umbrella review aimed to synthesize evidence from systematic reviews and meta-analyses on the efficacy of AI-driven exercise prescription for clinical outcomes in adults with chronic diseases, to appraise the methodological quality of the included reviews, and to quantify primary study overlap.

Methods:

We followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 and registered the review with PROSPERO (CRD420261345631). We included systematic reviews evaluating AI-based exercise prescription—using machine learning, deep learning, reinforcement learning, natural language processing, or generative AI—in adults with at least 1 diagnosed chronic noncommunicable disease vs usual care, no intervention, or conventional prescription. Search strategies for databases—PubMed, Embase, Web of Science, Cochrane Library, and IEEE Xplore—were drafted and peer-reviewed using the Peer Review of Electronic Search Strategies (PRESS) checklist and adapted to each platform’s syntax; databases were searched from inception to December 31, 2025 (all of which were last searched on that date). Methodological quality was assessed with AMSTAR 2 (A Measurement Tool to Assess Systematic Reviews 2), applied per the specification (item 2: protocol prespecification; item 4: search comprehensiveness). Certainty of evidence was rated using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach. Overlap was quantified as the corrected covered area (CCA). Reanalysis used random-effects models with robust variance estimation to accommodate correlated review-level estimates.

Results:

Eighteen reviews were included, comprising 232 study inclusions (187 unique primary studies). The most studied conditions were type 2 diabetes mellitus (n=6), cardiovascular disease (n=6), cancer (n=4), chronic obstructive pulmonary disease (n=3), and hypertension (n=3). The CCA was 1.4% (slight overlap). AI-based prescription was associated with reduced glycated hemoglobin (standardized mean difference [SMD] −0.36, 95% CI −0.46 to −0.25) and systolic blood pressure (WMD −5.23 mm Hg, 95% CI −7.68 to −2.78), and improved cardiorespiratory fitness (standardized mean difference 0.47, 95% CI 0.24 to 0.70). AMSTAR 2 rated 3 (16.7%) reviews as high, 8 (44.4%) as moderate, 6 (33.3%) as low, and 1 (5.6%) as critically low. GRADE certainty was moderate for these 3 outcomes and low or very low for the remainder. No outcome met the prespecified 10-estimate threshold for publication bias testing.

Conclusions:

The evidence is limited by moderate-to-low review quality, unexplained heterogeneity, absence of adverse event surveillance, and dependence among overlapping reviews that constrains precision. Within these limits, AI-based exercise prescription shows clinically relevant benefits across several chronic disease domains, most consistently for glycemic and blood pressure outcomes. Standardized algorithm reporting, adequately powered head-to-head trials, and prospective safety monitoring are research priorities.


 Citation

Please cite as:

Wang YT, Liu TC

AI-Based Exercise Prescription for Chronic Disease Management: Umbrella Review of Systematic Reviews and Meta-Analyses

DOI: 10.2196/105514

URL: https://preprints.jmir.org/preprint/105514

Download PDF


Request queued. Please wait while the file is being generated. It may take some time.

© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.