Training load prior to injury in professional Rugby League players: Analysing injury risk with machine learning

Abstract

This study explores the application of Global Positioning System tracking data from field training sessions and supervised machine learning algorithms for predicting injury risk of players across a single National Rugby League season. Previous work across a range of sporting codes has demonstrated associations between training loads and increased incidence of injury in professional athletes. Most of the work conducted has applied a reductionist approach, identifying training load characteristics as risk factors using generalised models to show population trends. This study demonstrates promising results by applying processing techniques and machine learning algorithms to analyse the injury risk associated with complex training load patterns. The accuracy of the algorithms are investigated along with the importance of training load predictors and data window sizes.