Anterior cruciate ligament (ACL) injuries remain a major concern in football, particularly among young athletes. Despite extensive research on biomechanical risk factors, identifying meaningful movement patterns within large datasets remains a significant challenge. This newly published study explores the application of artificial intelligence and unsupervised machine learning to characterize biomechanical phenotypes during a 90° change-of-direction task. The analysis included more than 6,000 trials performed by over 1,000 football players participating in the CutTheACL project.

By integrating kinematic, kinetic, and movement-quality variables, the machine learning algorithm identified four distinct movement phenotypes, each characterized by specific combinations of force production and neuromuscular control. These profiles ranged from athletes exhibiting low force generation and poor movement control to those demonstrating high force production with more favorable movement strategies.

The findings highlight the potential of data-driven approaches to reveal hidden biomechanical patterns that may not be detectable through conventional analyses. Such insights could contribute to more individualized screening procedures and targeted ACL injury prevention strategies in football. As artificial intelligence becomes increasingly integrated into sports medicine research, the identification of biomechanical phenotypes may represent an important step toward more personalized injury prevention programs.

The full article provides detailed methodology, statistical analyses, and discussion of the potential clinical and practical implications of these findings.