基於AI人體姿態辨識技術之帕金森氏症患者居家復健系統設計之研究

A STUDY ON THE DESIGN OF A HOME REHABILITATION SYSTEM FOR PARKINSON'S DISEASE PATIENTS BASED ON AI HUMAN POSE RECOGNITION TECHNOLOGY

李采姈(Tsai-Ling Li) ； 吳峻佑(Jun-Yo Wu) ； 黃晏亭(Yen-Ting Huang) ； 廖芸笙(Yun-Sheng Liao) ； 陳柏全(Bo-Cyuan Chen) ； 黃景晟(Ching-Cheng Huang) ； 余仁朋(Jen-Perng Yu) ； 朱美珍(Mei-Chen Chu)

《資訊電子學刊》 12卷2期 (2025/01) Pp. 151-171

摘要

隨著台灣的老年人口快速增長，帕金森氏症在老年族群中的發生率逐年提高，也越來越受到社會的關注。帕金森氏症是一種退化性神經疾病，會影響患者的生活品質和自主活動能力。儘管目前醫學上尚無法完全治癒，但藥物治療和運動復健可以有效延緩病情惡化。近年來，研究發現舞蹈治療也有助於增進帕金森氏症患者的運動能力。而運用日新月異的資訊科技技術，輔助帕金森氏症患者自主復健，提升患者復健的便利性與正確性是很值得探討的議題。本研究以AI人體姿態辨識技術為核心，結合機器學習、深度學習、OpenCV和Mediapipe等技術，建置一套輔助帕金森氏症患者之互動式居家自主復健系統。本研究除了包含基礎復健動作，還融入了舞蹈元素，不僅突破了傳統單調的運動模式，增添了趣味性，並提升患者的復健體驗，進而激發患者持續復健的動機與意願。系統能辨識患者的復健姿態，並即時比對動作的正確性，對錯誤姿勢提供修正指引，以確保復健動作的準確性。患者自主復健的紀錄和相關數據資訊都存取於資料庫中，提供復健師定期追蹤及觀察患者的復健狀況，以利給予患者最適當的復健規劃。

關鍵字

帕金森氏症 ； 肢體復健 ； 姿態辨識 ； 深度學習 ； TensorFlow ； OpenCV

並列摘要

As Taiwan's elderly population continues to grow rapidly, Parkinson's disease-primarily affecting older adults - has emerged as a pressing public health issue. This degenerative neurological disorder significantly impacts patients' quality of life and ability to perform daily activities. While there is currently no complete cure, medications and physical rehabilitation have shown to be effective in managing symptoms and slowing disease progression. Recent studies also indicate that dance therapy can enhance motor skills in individuals with Parkinson's. Harnessing advanced information technology to facilitate self-guided rehabilitation for Parkinson's patients presents an important opportunity to improve both the convenience and accuracy of their rehabilitation process. This study aims to develop an interactive, home-based rehabilitation system for Parkinson's patients, leveraging AI-driven human pose recognition alongside machine learning, deep learning, OpenCV, and Mediapipe. By integrating fundamental rehabilitation exercises with dance therapy, the system offers an engaging alternative to traditional exercises, enriching the rehabilitation experience and motivating patients to continue their routines. The system monitors patients' movements, assesses the accuracy of each rehabilitation action, and provides real-time corrective feedback to ensure precise execution. Rehabilitation records and relevant data are stored in a database, enabling therapists to track patient progress, review insights regularly, and create personalized rehabilitation plans that best support each patient's needs.

並列關鍵字

Parkinson's disease ； Physical Rehabilitation ； Pose Recognition ； Deep Learning ； TensorFlow ； OpenCV