Building Personalized Athlete Fatigue Prediction and Injury Risk Early Warning Model Using Wearable Sensors and Federated Learning

Abstract

Athlete fatigue accumulation and injury risk management are critical issues in competitive sports. Wearable sensor technology embedded in intelligent textile carriers enables real-time collection of multidimensional physiological and kinematic data, providing continuous and comfortable support for fatigue monitoring. Federated learning, as a distributed machine learning paradigm, enables collaborative modeling across multiple institutions while protecting athlete privacy. By integrating wearable sensing and federated learning, this study constructs personalized prediction models that dynamically assess fatigue levels and injury risk grades for early warning and intervention. The technical approach addresses the subjectivity and lag of traditional assessment methods and provides an intelligent solution for scientific training management and injury prevention. At the current stage, the system still has limitations in long-term chronic injury prediction, requiring further optimization of temporal modeling and personalized adaptation algorithms. The study also expands the application of intelligent textiles in sports monitoring, where wearable antennas, wireless telemetry, and electromagnetic-compatible sensor integration are important for stable data transmission in high-intensity movement environments.