3D Machine vision and deep learning for enabling automated and sustainable assistive physiotherapy

Smith, Lyndon; Boyd, Stephen; Bhatta, Devaki; Smith, Melvyn

Abstract

The significant beneficial effects of physiotherapy as a treatment for a wide range of medical conditions, is well known. However, the challenge for already stressed healthcare systems to provide effective physiotherapy to an aging population, that has increasing requirements and often limited mobility, is formidable. Further pressure on provision has arisen as a consequence of the COVID-19 pandemic and climate change concerns over carbon emissions associated with travel of the patient and/or physiotherapist to treatment centers. The situation is particularly acute in relation to frail/elderly patients, due to increased demand placed on services by the pandemic, the risk to the patient of infection or of suffering a fall, and the relatively high fiscal and environmental costs associated with face-to-face care. These factors provide strong motivators for effective provision of remote healthcare methods; however, before automated physiotherapy systems can become widespread, they need to be able to appraise patients and monitor their responses to recommended physiotherapy regimes, with an accuracy similar to that of human physiotherapists and at an acceptable level of cost. This paper outlines a novel remote healthcare package that aims to provide elderly persons with accurate, accessible, and cost-effective physiotherapy in their own homes. The approach makes use of emerging state-of-the-art machine vision (incorporating 3D data to improve accuracy) and advanced machine learning techniques, for accurately recovering a patient's joint and limb positions during exercises, with the ultimate aim of enabling automated assessment of patient exercise concordance. This is considered to be a first step toward a system that can assess patients and monitor their progress, as part of an automated approach to physiotherapy that can offer significant access, environmental and cost benefits; thereby assisting with on-going sustainability of healthcare provision.

Citation

Smith, L., Boyd, S., Bhatta, D., & Smith, M. (2024). 3D Machine vision and deep learning for enabling automated and sustainable assistive physiotherapy. In 2023 Congress in Computer Science, Computer Engineering, & Applied Computing (CSCE) (1247-1253). https://doi.org/10.1109/CSCE60160.2023.00209

• Conference Name: 2023 Congress in Computer Science, Computer Engineering, & Applied Computing (CSCE)
• Conference Location: Las Vegas, USA
• Start Date: Jul 24, 2023
• End Date: Jun 27, 2023
• Acceptance Date: Mar 15, 2023
• Online Publication Date: Apr 9, 2024
• Publication Date: Apr 9, 2024
• Deposit Date: Jun 29, 2023
• Publicly Available Date: Apr 10, 2026
• Publisher: Institute of Electrical and Electronics Engineers (IEEE)
• Pages: 1247-1253
• Book Title: 2023 Congress in Computer Science, Computer Engineering, & Applied Computing (CSCE)
• ISBN: 9798350327601
• DOI: https://doi.org/10.1109/CSCE60160.2023.00209
• Public URL: https://uwe-repository.worktribe.com/output/10895791