Preventing Injuries in Triathletes with Machine Learning using Wearable Sensor Data

Wearable devices now allow endurance athletes to collect continuous physiological data that reveals critical insights into training responses and emerging injury risk. However, interpreting this abundant data is not always trivial. Integrating it with predictive machine learning systems presents a major opportunity to advance injury prevention. A major restriction are privacy restrictions in health and sports science domains, impeding large-scale access to athlete data, essential for training high-quality predictive models. This limitation creates a bottleneck in deploying effective early-warning systems for injury prevention. This thesis proposes a novel solution through the creation of a synthetic data generation framework.