Multimodal AI Predicts Tennis Player Injury Risk

A new multimodal machine learning framework, PART (Predictive Athlete Readiness framework for Tennis), has been developed to assess both performance and injury risk in tennis players. Unlike previous approaches that often rely solely on subjective observations, PART integrates a wide array of data sources. These include physiological metrics, training and match data, sleep patterns from wearable devices, daily self-reported information, jump assessments, and detailed motion analysis extracted from match play videos. PART processes this diverse data to capture four key characteristics of tennis players: overall wellness, injury risk, physical capability, and playing style. By combining these characteristics through supervised learning, the framework provides a holistic assessment of an athlete's condition and offers advanced forecasts of specific body areas at risk, such as elbows or knees. Evaluated with data from nine collegiate tennis players, PART demonstrated strong performance in predicting both overall wellness and injury risk, also showing promise for recreational players who often suffer technique-related injuries.