In professional sports, optimal performance requires a balance between training and subsequent recovery. To follow-up on this balance, it is important to monitor training load, symptoms of fatigue and changes in performance. At present, performance is mostly monitored via standardized tests, which require the athlete to go all out in a controlled environment with expensive measuring materials. As these tests might disrupt the athlete’s training schedule and require a specific infrastructure, the first objective of this project is to derive approximate performance measures that are observable from training and wearable data and learn which additional parameters influence these proxy performance measures. The latter is important to move from monitoring performance to robustly predicting performance.
Though prediction can already aid in actively influencing the course of performance, the second objective of the project is to go beyond predictive modelling. The research aims at developing a model which can provide athletes with concrete actions to undertake given a desired performance in the future, i.e., a prescriptive model. The research will initially focus on cycling. However, results will be transferable to other endurance sports.