Computer-Based Machine Learning Model for Supporting and Healing ACL Injury for Athletes

Abstract

Physical rehabilitation exercises play an extraordinary role in healthcare field, s pecifically fo r pa tients wi th ACL (anterior cruciate ligament) injury, aiding in the recovery and improvement of various physical conditions. In this context, we propose a powerful classification a lgorithm for categorizing five exercises for patients with ACL injury in their first p hase of rehabilitation journey. In this paper, we utilized the Mediapipe framework to extract six key joints (right hip, left hip, right knee, left knee, right ankle, left ankle), and subsequently derived statistical measures for each joint. Furthermore, these statistical measures serve as input for our implemented machine learning models, which are Support Vector Machine (SVM), and Random Forest to classify the exercises based on their coordinates extracted. Then we get the assessment of each exercise performed by the patient to provide feedback to patients by the physiotherapists. Moreover, this will allow doctors to monitor patients' progress more efficient through their journey in their rehab. The proposed approach of Support Vector Machine (SVM) model achieved 96.61% accuracy and Random Forest model achieved 97.52%. These results reveal how practical our approach is in making the correct classification of rehabilitative exercise, thus a valuable resource for enhancing the recovery of the patients, aiding physiotherapists in giving more informed feedback