Innovative machine learning approach for analysing biomechanical factors in running-related injuries

Running-related injuries are a significant concern for recreational and competitive athletes, often resulting from complex biomechanical interactions. Traditional injury assessment methods are limited in their ability to capture dynamic, real-time data, necessitating the need for more advanced predictive tools. This study proposes an innovative machine-learning approach to predict running-related injuries by analyzing biomechanical data collected from 84 active runners. The data included joint angles, ground reaction forces, stride length, muscle activation, and foot pressure, captured through wearable sensors during laboratory-controlled and outdoor running sessions. An ensemble model combining Gradient-Boosted Decision Trees (GBDT), Long Short-Term Memory (LSTM) networks, and Support Vector Machines (SVM) was developed to predict injury risk. The results indicate that ground reaction force, foot pressure, and stride length were the most significant predictors of injury. The proposed ensemble model achieved an accuracy of 88.37%, outperforming individual models such as GBDT (83.74%) and LSTM (81.29%). The findings suggest that integrating machine learning techniques with biomechanical analysis can significantly enhance the prediction and prevention of running-related injuries. This research offers valuable insights into developing personalized injury prevention strategies, potentially reducing injury occurrence among athletes.