本研究以AI人體姿態辨識為主軸，結合機器學習、OpenCV及Google Mediapipe AI Pose肢體辨識、語音辨識、D3.js等技術、建置一套協助肌少症患者之互動式復健輔助系統。本研究將復健運動，透過偵測肢體特徵之各定位點及復健師指示之關鍵動作來進行自我復健，並提供修正指示，協助患者改善復健動作。另外，本研究亦開發復健師系統，將復健紀錄及復健影片視覺化呈現至系統中且可訂定患者復健目標，以利復健師追蹤並觀察患者復健的過程。本研究完成之系統，協助肌少症患者能夠居家隨時進行復健，再配合系統指示來進行復健動作改善，提高患者復健成效之餘，也將復健方式推行得更加智慧且便捷。

In this study, we focused on AI human body posture recognition, combined with machine learning, OpenCV and Google Mediapipe AI Pose body recognition, speech recognition, D3.js and other technologies, etc. to develop an interactive rehabilitation assistance system for sarcopenia patients. The study, rehabilitation exercise will be carried out by detecting the positioning points of limb features and the key movements instructed by the rehabilitator to carry out self-rehabilitation, and provide correction instructions to help patients improve rehabilitation movements. In addition, this study also developed a system for rehabilitators, which visualized the rehabilitation records and videos into the system and set the patient's rehabilitation goals, so that the rehabilitators could track and observe the patient's rehabilitation process. The system completed in this study helps sarcopenia patients to carry out rehabilitation at home at any time, and then cooperates with the system instructions to improve rehabilitation movements, which not only improves the rehabilitation effect of patients, but also promotes the rehabilitation method to be more intelligent and convenient.