Sepsis is a major cause of morbidity and mortality in children and young people. Early detection is critical to improve treatment outcomes, but current methods to predict sepsis onset are limited. PHEMS aims to address this by developing an algorithm to predict sepsis in pediatric intensive care units (PICUs). To support the development of this algorithm, PHEMS held an online hackathon using real-world PICU data. Participants were asked to predict sepsis six hours prior to clinical onset. The hackathon provided a valuable testing ground for evaluating different machine learning strategies. The results will help ensure that the final PHEMS sepsis prediction algorithm is both high performing and privacy-compliant.
