Interpretable Machine Learning in Athletics for Injury Risk Prediction

Collegiate athletics have been very competitive in bringing in talent and competing at the highest level. Coaches and the training team aim for the best team performance during the season while keeping the athletes healthy at their prime. Data analytics has provided the analytical frameworks/tools to gather and analyze athletes’ data and assist coaches in making game-time decisions as well as in designing and analyzing and customizing various training regimens. In this chapter, we will overview some of the key machine learning (ML) and computer vision techniques in the context of sports data analytics (SDA) that have been employed in collegiate athletics. These techniques can be used for optimization and streamlining processes such as designing training plans for athletes, formulating match strategies, analyzing and improving athlete and team performance, predicting the risk of injury to an athlete, athlete profiling, and team selection. Our case study covers injury risk prediction for a Women’s Division I basketball team using video data. We build ML models whose predictions/decisions are interpretable (XAI) to human experts.