383 Artificial Neural Network Modeling of Outcomes in Subarachnoid Hemorrhage

Abstract

INTRODUCTION: Aneurysmal subarachnoid hemorrhage is a devastating neurological condition which requires complex neurocritical care and neurosurgical decision-making yet results in remarkably variable outcomes. Robust methods of identifying patients who may need operative and critical care interventions, such as those who develop hydrocephalus requiring long-term shunting or cerebral vasospasm, or predicting patients who may go on to have long-term disability or worsened neurological outcomes are lacking. METHODS: Data used to train our models included demographic data (gender, age and race), history of medical conditions (hypertension, diabetes, coronary artery disease, hyperlipidemia, tobacco use), Hunt and Hess scale, ventriculostomy data (external ventricular drain (EVD) output, EVD level, EVD age), and cerebrospinal fluid (CSF) lab values (CSF protein and glucose levels). The models were trained on a subpopulation of 85% of subjects and then individually validated in a training-naive subset of 15% of subjects with high accuracy. RESULTS: Using this data, we have generated artificial neural network classifiers with hyperparameter tuning which predict the need for long-term cerebrospinal fluid (CSF) diversion via placement of a shunt (receiver operating characteristic area under the curve) (ROC AUC) = 0.8312), vasospasm (ROC AUC = 0.8058), and the categorical outcome of patient disposition (model accuracy = 0.7786). CONCLUSIONS: We hope the use of artificial intelligence and machine learning techniques will continue to demonstrate power in predicting complex medical outcomes and ultimately help neurosurgeons and neuro-critical care personnel prognosticate and provide appropriate and timely treatment.