AI-based precision prognostics and therapy personalization for childhood brain tumors

Accurate prediction of patient outcomes remains a critical challenge in oncology, in particular for childhood brain tumors, in which treatment decisions need to carefully balance risk of tumor recurrence and death against severe therapy-induced sequelae. Here, we developed MANTIS (methylation profiling and artificial neural networks for time-dependent individualized survival predictions), a precision artificial intelligence (AI) tool for medulloblastoma and ependymoma outcome prediction built on genome-wide DNA methylation array data from 2,786 patients. MANTIS generates individualized survival probabilities over a 10-year period and achieves concordance indices as high as 0.80 ± 0.07, significantly exceeding the predictive accuracy of current clinical indicators. Extending MANTIS to estimate individualized treatment effects identifies ~60% of patients with average-risk medulloblastoma without a significant survival disadvantage (P = 0.683) when treated with less intensive craniospinal irradiation doses, which is associated with improved neurocognitive outcomes. In conclusion, our study establishes a foundation for AI-based precision prognostics and therapy personalization in childhood brain tumors and offers a blueprint for similar initiatives in other cancer types.