Artificial intelligence applied to performance analysis and injury prevention in team sports: a narrative review

Abstract

Artificial intelligence (AI) is transforming sports sciences, offering new tools for performance analysis and injury prevention in team sports. The aim of this narrative review was to synthesize the available scientific evidence on AI applications in team sports, identifying the main techniques used, their areas of application, and the associated ethical and methodological challenges. A literature search was conducted in specialized databases, selecting studies published between 2020 and 2026. The results indicate that machine learning, particularly tree-based methods (Random Forest, XGBoost), artificial neural networks, and deep learning, have demonstrated promising accuracy in injury risk prediction and performance optimization. Computer vision facilitates automated tactical analysis and markerless biomechanical assessment. Wearable sensor-based monitoring systems enable continuous tracking of workload and wellness indicators. However, significant limitations persist related to methodological heterogeneity, small sample sizes, lack of data standardization, algorithmic opacity, and ethical considerations regarding athlete privacy and autonomy. It is concluded that, while AI offers considerable potential to improve performance and reduce injury incidence in team sports, its effective and responsible implementation requires addressing these challenges through longitudinal studies, explainable models, and robust ethical frameworks.