본 연구는 RGB 카메라 기반 자세 추정 알고리즘을 활용하여 어깨 운동 중 보상 움직임을 실시간으로 검출하고, 관절 각도를 측정하는 디지털 트윈 재활 시스템을 구현하였다. 게임 인터페이스와 통합하여 실시간 바이오피드백을 제공하며, 파일럿 테스트 결과 사용자 만족도 4.3/5.0점을 기록하였다. 본 연구는 비접촉식 영상 분석 기술을 통한 접근성 높은 재활 게임 플랫폼의 가능성을 제시한다.

Patients with shoulder joint injuries exhibit excessive activation of trunk muscles to compensate for limited range of motion. This study implements a digital twin system that utilizes video analysis technology to detect compensatory movements during shoulder exercises in real-time and applies this to functional rehabilitation games. Using RGB camera-based pose estimation algorithms, we quantified compensatory movements of the spine and trunk during shoulder elevation, and developed a system that evaluates exercise effectiveness by measuring key joint angles including flexion, extension, abduction, and adduction. Demonstration content was developed by integrating this with a game interface to provide real-time biofeedback. Pilot testing with one elderly shoulder rehabilitative exercise patient demonstrated that the system detected arm compensatory movements with improved accuracy, achieving a user satisfaction score of 4.3 out of 5.0. This research demonstrates the potential of an accessible rehabilitation game platform utilizing non-contact video analysis and digital twin technology

ABSTRACT
요약
1. 서론
2.연구방법
    2.1 시스템 설계
    2.2 보상운동 검출알고리즘 구현
    2.3 보상운동 검출알고리즘 구현
3. 파일럿 케이스 실험
    3.1 굴곡 각도(Flexion Angle)
    3.2 외전 각도(Abduction Angle)
    3.3 몸통 전방 기울임(Trunk ForwardLean)
    3.4 흉추 과신전(Thoracic Extension)
    3.5 몸통 회전(Trunk Rotation)
    3.6 좌우 어깨 비교 분석
    3.7 대조군 분석
    3.8 기능성 게임 모듈으로서의 시스템 성능평가
4. 결론
참고문헌

• 장태수(서울아산병원 의공학연구센터) | Taesoo Jang
• 유인화(메디온 AI운동의과학연구소) | Inhwa Yoo Corresponding author