| Designing A Video-Based Artificial Intelligence Model for Personalizing Corrective Exercises in Adolescents With Scoliosis |
| Paper ID : 1109-SPORTCONGRESS (R2) |
| Authors |
|
Vahid Afshoon *1, Ghazal Vazirian2 1Department of Physical Education, Shushtar Branch, Islamic Azad University, Shushtar, Iran. 2Master's degree in Motor Behavior-Motion Developmental Orientation University of Tehran |
| Abstract |
| Scoliosis is a common spinal deformity among adolescents, often requiring long-term corrective exercise programs to prevent progression and improve posture. However, conventional rehabilitation approaches lack personalization and rely heavily on manual assessment, limiting their effectiveness. This study presents a novel video-based artificial intelligence (AI) model designed to analyze postural and movement patterns in adolescents with idiopathic scoliosis and generate individualized corrective exercise plans. A total of 45 participants aged 12–18 were recruited and underwent an 8-week intervention guided by the AI-generated programs. The model utilized computer vision techniques to extract skeletal keypoints from video recordings and applied machine learning algorithms to detect deviations and prescribe tailored exercises. Results showed a significant reduction in Cobb angle (mean decrease of 4.7°, p < 0.01), improved shoulder symmetry, and enhanced pelvic stability. The AI system achieved 92.3% accuracy in posture detection and received high clinical approval from physiotherapists. This research highlights the potential of AI-powered video analysis as a scalable, non-invasive, and effective solution for personalized scoliosis rehabilitation, bridging the gap between technology and therapeutic practice. |
| Keywords |
| Exercise, Muscle Mass, Strength (Example) |
| Status: Abstract Accepted (Poster Presentation) |
