Real-Time Monitoring of Rehabilitation Outcomes for Track and Field Athletes Using Artificial Intelligence Wearable Devices

Wearable devices and human-computer interaction technologies have revolutionized rehabilitation monitoring by enabling real-time, objective, and dynamic assessments. This study integrates wearable sensors, including triaxial accelerometers, gyroscopes, and heart rate monitors, with machine learning algorithms to analyze rehabilitation progress in track and field athletes. Advanced preprocessing techniques, such as noise reduction and feature normalization, are applied to ensure data quality. The system employs Gradient Boosting Machine (GBM), Random Forest (RF), and Support Vector Machine (SVM) models to classify recovery stages and detect anomalies. Experimental results show that the GBM model achieves the highest performance, with an accuracy of $\mathbf{9 4. 3 \%}$ and an AUC-ROC of $\mathbf{0. 9 6}$, surpassing RF and SVM. Feature importance analysis highlights lower-limb acceleration and heart rate variability as key indicators of recovery, underlining the system’s capability to provide actionable and personalized insights. This data-driven approach addresses limitations in traditional rehabilitation monitoring methods, which often rely on subjective assessments and static evaluations. By combining wearable technology with machine learning, the proposed system enhances recovery monitoring, offering precise and reliable support for athletes and medical professionals. The results confirm the effectiveness of artificial intelligence in advancing rehabilitation processes and provide a foundation for practical applications in sports science and injury recovery.