Enabling Training Personalization by Predicting the Sessioon Rate of Perceived Exertion

In order to achieve optimal gains from training while reducing the risk of negative effects, athletes must perceive an ideal Internal Training Load. This load is invoked by the training content and individual characteristics. Training sessions are often team-based, although each athlete has his/her own characteristics, making it hard for coaching staff to optimize the training load for each athlete individually. In this paper, we propose a methodology for a decision support system that predicts three different sRPE scores: an average group score and an individualized score prior to and after training. This system can be used to personalize training sessions and to follow-up the training in a real-time fashion. We report results on a dataset collected from association football (soccer) players of a Belgian club. The reported results are better than current state-of-the-art from literature in the Australian football domain, with a smaller dataset and less invasive features.