Machine learning applications in sport: a scoping review

Machine learning (ML) applications continue to grow in popularity across the sport industry, offering new opportunities for performance enhancement, injury prevention, and decision-making. The present scoping review examined the landscape of ML applications in sport by analyzing 270 peer-reviewed studies published between 2002 and 2024. ML was applied across 12 broad subject areas, with computer science, biomechanics, and sport psychology emerging as the most common domains of application. Key applications included action recognition, injury prediction/prevention, and athlete selection/talent identification. While ML models have demonstrated promising accuracy, their practical utility was often limited by issues of data quality, interpretability, and accessibility for end users such as athletes, coaches, and sport interest-holders. Given the issues surrounding the practical usability of ML, the ultimate goal of ML should be to support - not replace - human expertise, as its integration may enhance sport and athlete experience at all levels of engagement and development.