SURG-41. MACHINE LEARNING PREDICTS THE INFLUENCE OF FRAILTY ON THE CLINICAL MANAGEMENT OF NEUROFIBROMATOSIS TYPE 1: A MIXED-EFFECTS MODELING STUDY USING THE NATIONWIDE READMISSIONS DATABASE

Abstract Recent literature has highlighted the association of frailty with significantly higher rates morbidity and mortality in patients CNS neoplasms. However, there is a paucity research regarding effects as it relates to neurocutaneous disorders, namely NF1. In this study we evaluated role NF1 compared its predictive utility that Elixhauser Comorbidity Index (ECI). Namely, used Nationwide Readmissions Database from 2016 2017 select for those diagnosis who underwent neurosurgical resection an intracranial tumor. Patient was queried through use Johns Hopkins Adjusted Clinical Groups (JHACG) frailty-defining indicator. ECI scores were collected these measurement comorbidities. Propensity score matching performed age, sex, ECI, insurance type, median income by ZIP code, yielded 60 frail non-frail patients. Receiver operating characteristic (ROC) curves created relevant complications, including mortality, non-routine discharge, cost top quartile, length stay [LOS] readmission, using comorbidity indices predictor values. The area under curve (AUC) each ROC served proxy model performance. Following groups, still had increased average hospital ($85,441.67±$59,201.09) ($49,321.77±$50,705.80)(p=0.010). Similar trends also found LOS between (23.1±14.2 days) (10.7±10.5 days)(p=0.0020). For complication interest, plotted showed performing well combination Frailty+ECI (p >0.05). (AUC: 0.924)outperformed only 0.842, p=0.027). These findings suggest efficacious predicting complications high lesions associated