Previously submitted to: JMIR Medical Informatics (no longer under consideration since Feb 06, 2026)
Date Submitted: Jan 6, 2025
Warning: This is an author submission that is not peer-reviewed or edited.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.
Smart Stroke Rehabilitation and Scoring Framework using Artificial Intelligence and Smartphone
ABSTRACT
Background:
Post-stroke upper extremity assessment is crucial for monitoring stroke recovery, yet it often remains inaccessible to patients in underserved communities. Existing artificial intelligence-based assessments frequently depend on expensive, cumbersome hardware and show limited classification performance and automation, restricting their scalability and accessibility.
Objective:
To address these challenges, we propose an end-to-end deep-learning-driven framework for accurate and automated scoring of the Fugl-Meyer Assessment - Upper Extremity (FMA-UE) using video data captured by smartphones, aiming to bboth scalability and accuracy in post-stroke recovery assessment.
Methods:
The proposed framework consists of two key modules: First, a top-down joint location (JL) module detects 2D body and hand joint coordinates from smartphone videos. These joint locations are then processed through a novel centering and sampling pipeline, yielding a structured matrix of coordinates. Next, a classification module employs a 3D heatmap mechanism to extract spatiotemporal features from the coordinates via 3D convolution. The framework was evaluated on real-world data from stroke patients, focusing on 14 out of 30 FMA-UE items. Accuracy and F1-scores were used to to compare the proposed framework against existing methods.
Results:
Our model achieved an accuracy ranging from 82.6% to 95.3% across the evaluated FMA-UE items, with an average accuracy of 88.6% and an F1-score of 0.883. These results demonstrate the model's effectiveness in various assessment tasks. Compared to existing methods, the proposed framework exhibited superior performance, providing accurate, automated, and scalable assessment.
Conclusions:
We demonstrate the feasibility of a smartphone-based, deep-learning-powered system for FMA-UE autoscoring in post-stroke recovery assessment. By offering high accuracy alongside ease of use and scalability, this framework presents a promising solution for both clinical and at-home rehabilitation, potentially leading to broader adoption in stroke recovery diagnostics.
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.