A multicenter study demonstrates that deep transfer learning (DTL) algorithms using preimplant MRI can predict spoken language development in children with cochlear implants with over 92% accuracy, significantly outperforming traditional machine learning models and clinical variables.