The aim of this work is to provide an overview of the modern applications of artificial intelligence (AI) in professional sports, using football as an example. We will discuss modern approaches to injury prediction and the requirements for implementing these systems in a club setting. The methodology includes a scoping review of publications from 2015 to 2025 in the fields of sports medicine, sports science, and computer vision, as well as an analysis of industry standards for data quality. As a result, key classes of AI tasks have been systematized, including tracking, scouting, and decision support by headquarters. Typical data and model architectures have been identified, as well as critical sources of information bias and leakage. Requirements for validation and implementation have also been highlighted. It has been demonstrated that injury risk models in football show potential in prospective studies with high-quality «time-loss» labeling and strict time partitioning. However, their practical value depends not only on AUC, but also on calibration, resistance to data drift, and integration into the training process and recovery measures. The practical significance of the results is in the development of a practical framework for creating an «AI club system». This includes standardizing data, creating an ethical and legal framework, selecting metrics, establishing a monitoring protocol, and developing a set of preventive measures supported by models. The value of this work lies in its focus on sports scientists, analysts, and coaching staff, as well as its emphasis on the reproducibility and risk management of AI implementation.
Kronfel'd et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: