Predictive modelling for sports performance improvement and injury prevention

Abstract

Sport-related injuries occur due to a complex interaction of many internal and external risk factors gathered in a pattern of either positive adaptation (increased fitness), or negative adaptation (injury). The repetitive nature of the high-speed full-body pitching movement exposes the pitcher’s elbow to high loads. This thesis illustrates a novel approach to individualised injury risk prediction that accounts for the dynamics of the injury development process. The integration of advanced monitoring techniques plays an important role in the pursuit of high-level sports performance. The utilization of wearable sensors serves that purpose. It allows continuous athlete assessment and provides feedback on the relevant health and performance metrics in real–time. The methods established in the thesis offer solutions for dealing with different quality, time scale and hierarchical data structures collected with high-end wearable sensors, self-reported questionnaires and motion capture systems. Integration of the available data from different sources and implementation of the statistical models that can translate them to the relevant outcome provides actionable insights for performance improvement and injury prevention. Adding these statistical methods is a chance for training and injury-prevention programs to continue to improve.