Currently submitted to: JMIR Formative Research
Date Submitted: Mar 21, 2026
Open Peer Review Period: Mar 22, 2026 - May 17, 2026
(currently open for review)
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
A Pilot Feasibility Study of an AI-Based Video Analysis System for Physiotherapy Assessment in Patients With Hemiplegic Stroke
ABSTRACT
Background:
Traditional physiotherapy assessments are performed manually using standardized scales, which are time-consuming and subject to rater variability. Artificial intelligence (AI) offers potential for objective, efficient, and standardized rehabilitation evaluation.
Objective:
This study aimed to evaluate the feasibility of an AI-based video analysis system for physiotherapy assessment in patients with hemiplegic stroke and to explore its agreement with conventional clinical assessments.
Methods:
A mixed-methods design was employed, including literature review, modeling analysis, case studies, and semi-structured interviews. An AI framework was developed to analyze patient videos performing predefined actions, extracting data for Activities of Daily Living (ADL), balance, and range of motion (ROM). AI-derived results were compared with manual assessments. Agreement was quantified using statistical tests and intraclass correlation coefficients (ICC), and discrepancies were explored through triangulation.
Results:
Ten stroke patients were assessed, yielding 30 paired comparisons. For balance, no significant differences were observed between AI and manual methods (t-test, P > .05). ADL evaluation showed good to excellent agreement (ICC = .92 for total score; ICC = .66 for ADL-6 subscale). ROM assessment demonstrated large discrepancies, with mean errors exceeding clinically acceptable thresholds, and was excluded from final analysis. The small sample size (n=10) limits generalizability. These findings provide preliminary evidence supporting the feasibility of AI-assisted assessment in selected domains.
Conclusions:
This pilot study provides preliminary evidence that AI-based video analysis can achieve comparable performance to manual assessment in balance and ADL evaluation of hemiplegic stroke patients. However, methodological limitations and insufficient accuracy in ROM assessment highlight the need for larger-scale validation and integration of external sensors before clinical application. Clinical Trial: Not applicable.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© 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.