Automated Platform for the Kinematic Assessment of Hand Motor Deficit in Patients with Neurological Diseases
摘要
This work presents an automated video analysis platform for assessing hand motor function in patients with stroke, multiple sclerosis, and Parkinson’s disease. Videos of patients performing hand exercises were processed with MediaPipe Hands to extract 21 keypoints per hand, from which kinematic metrics (distances, angles, velocities) and statistics were computed. These features were used in binary and multiclass classification models to identify pathology. A web application integrates the full workflow—video upload, automated analysis, and PDF report generation.