A machine learning-based life activity classification and qualitative intervention program to promote physical activity for people with disabilities

2023 Fiscal Year Final Research Report

• Project/Area Number: 20K11305
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 59020:Sports sciences-related
• Research Institution: The University of Electro-Communications
• Principal Investigator: Ohkawara Kazunori 電気通信大学, 大学院情報理工学研究科, 教授 (30631270)
• Co-Investigator(Kenkyū-buntansha): 稲山 貴代 長野県立大学, 健康発達学部, 教授 (50203211)
• Project Period (FY): 2020-04-01 – 2024-03-31
• Keywords: 機械学習 / 活動分類 / 障害者

Outline of Final Research Achievements

In this study, we attempted to examine the optimal algorithm for classifying physical activity based on acceleration values for spinal cord injured who use wheelchairs and visually impaired persons, using a random forest. As a result, although sufficient estimation accuracy could not be obtained for the 9-item classification, an average estimation accuracy of about 70% was obtained by 3 or 4 categories, which is a level of accuracy that can be expected for practical use in the future. In the development of an application for practical use, the acquired continuous acceleration data was processed on a smartphone, and the results of physical activity amount and activity classification were successfully visualized.

Free Research Field

応用健康科学

Academic Significance and Societal Importance of the Research Achievements

障害者は健常者に比べて生活習慣病への罹患リスクが高いといわれているにもかかわらず、その予防策の具体的な手段は提案されていない。本研究は、身体活動レベルや生活活動パターンを簡便かつ高い精度で評価し、その結果が可視化できることを示唆した。この成果は、身体活動の増加を促進する手段の提案につながるエビデンスとして位置付けられる。厚生労働省（2016年調査）によると18歳以上の在宅の身体障害者は400万人を超えており、地域社会の持続的発展へも貢献すると考えている。