Previously submitted to: JMIR mHealth and uHealth (no longer under consideration since Mar 01, 2023)
Date Submitted: Dec 7, 2022
Warning: This is an author submission that is not peer-reviewed or edited.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.
Smartphone App for patients with hemiplegia to support self-administered graded upper extremity exercise using deep learning model: Development and Usability Study
ABSTRACT
Background:
Regular self-administered exercise after stroke can improve upper limb function. For proper self-administered exercise, an appropriate guide is required to match the exercise level to the patient’s upper limb disability.
Objective:
This study aimed to develop a mobile health application to support self-administered graded upper limb exercises for patients with hemiplegia using deep learning model.
Methods:
To develop a patient-centered, solution-based mobile application for upper extremity exercises for hemiplegia, we used a 5-step design thinking process. The exercise programs in the application were safe and easy to follow in the sitting position. A deep learning model was proposed to automatically evaluate the patients’ upper limb function for recommendation of exercise according to each patient’s disability. The model captured videos of patients reaching and grasping objects on a table as input and predicted the exercise level. Potential end-users, including patients, therapists, and physicians, evaluated the prototype and provided feedback for the usability and quality of the application using a mobile application rating scale.
Results:
A total of six features, my information, evaluation section, daily exercise, list of exercises, my current exercise status, and notification, were incorporated into the mobile application. The application consisted of 58 exercises categorized as postural and balance, range of motion, strengthening, and task-oriented training exercises with three levels of difficulty. A deep-learning-based model was proposed for automated upper limb function evaluation. The overall accuracy of classification of exercised level was 84.58% and inference time was 9~10 seconds.
Conclusions:
We successfully developed a mobile application to support self-administered graded upper limb exercises for patients with hemiplegia based on individual ability, including automatic evaluation using a deep learning model.
Citation
The author of this paper has made a PDF available, but requires the user to login, or create an account.
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.