Nowadays, professional soccer players are individually monitored during training sessions and matches to both optimize their physical fitness and reduce their injury risk. The availability of wearable sensors allows their clubs to collect a vast amount of data every day. However, to leverage these data, actionable advise needs to be presented to the medical staff and the team manager. Currently, this is challenging for two reasons. First, manually analyzing these data for every player after every training session or match is practically not feasible. Second, the link between the external load (i.e., all activities performed by the players during a session or match) and the internal load (i.e., the psychophysiological stress experienced by the player) in professional soccer players is not yet fully understood. In this presentation I will present how machine learning techniques provide a data-driven approach to tackle both challenges.