MDB-11. AN ONLINE CALCULATOR USING MACHINE LEARNING FOR PREDICTION OF SURVIVAL IN PEDIATRIC PATIENTS WITH MEDULLOBLASTOMA

Abstract OBJECTIVE Medulloblastoma is the most common malignant intracranial tumor affecting pediatric population. Despite advancements in multimodal treatment over past 2 decades yielding a >75% 5-year survival rate, children who survive often have substantial neurological and cognitive sequelae. The authors aimed to identify risk factors develop clinically friendly online calculator for prognostic estimation medulloblastoma patients. METHODS Pediatric patients with histopathologically confirmed were extracted from Surveillance Epidemiology End Results database (2000-2019) split into training validation cohorts an 80:20 ratio. Cox proportional hazards model was used univariate multivariate predictors. Subsequently, those developed predict 2-, 5-, 10-year overall as well median months performance of determined by discrimination, calibration, decision curve analysis (DCA). RESULTS A total 1,739 met prespecified inclusion criteria. Fourteen variables, including age, sex, race, ethnicity, household income, county attribute, laterality, histology, anatomical location, grade, size, surgery status, radiotherapy, chemotherapy, included (https://spine.shinyapps.io/Peds_medullo/). concordance index 0.757 cohort 0.762 cohort, denoting useful predictive accuracy. Good agreement between predicted observed outcomes demonstrated calibration plots. DCA curves indicated that has good clinical benefit CONCLUSION An easy-to-use large established. Future efforts should focus on improving granularity population-based registries externally validating proposed calculator.