Sponsored by
Researchers at Mass General Brigham, in collaboration with Boston Children’s Hospital and Dana-Farber, developed an AI model using a method called temporal learning to predict tumor recurrence in children with gliomas. The model analyzes sequences of post-treatment scans to detect subtle changes over time, achieving prediction accuracy rates of 75% to 89%. The model's accuracy plateaued between 4 to 6 scans. It's worth noting that the study was made possible by access to nearly 4,000 scans from 715 pediatric patients, with imaging and clinical data access support from the Children’s Brain Tumor Network (CBTN).