Digital twin framework for injury risk prediction and management in competitive sports

This paper presents a digital twin (DT) framework aimed at predicting and managing injury risks in competitive sports. The proposed framework integrates biomechanical data, machine learning, and real-time monitoring to support athlete health and performance. DT technology enables detailed performance analysis and illness prediction through virtual athlete models. While previous studies have used wearable sensors and machine learning, they often lack cross-domain integration. This research addresses that gap by introducing a nine-stage DT development pipeline incorporating biomechanical data, PCA-based preprocessing, and musculoskeletal modelling. Validation through cross-validation and evolutionary data scenarios demonstrated the models robustness. XG Boost achieved the highest injury prediction accuracy (78%), with key predictors including hamstring force and muscle stiffness. Biomechanical simulations revealed stress patterns consistent with physiological behaviour, supporting clinical relevance. Comprehensive decision-support systems remain scarce. This work contributes toward safer, more personalised sports environments using a holistic, data-driven DT approach.