Development and Validation of Machine Learning Models for the Prediction of Blunt Cerebrovascular Injury in Children

Machine learning model using National Trauma Databank data (885,100 pediatric patients) achieves 94.4% sensitivity for detecting blunt cerebrovascular injury in children, significantly outperforming adult-derived Denver and Memphis criteria (13.4% sensitivity). Skull fractures, extremity fractures, and vertebral injuries were key predictive features.