401 Applying Machine Learning for Risk Stratification and Acute Clinical Outcome Prediction Amongst Aneurysmal Subarachnoid Hemorrhage Patients

Abstract

INTRODUCTION: Subarachnoid hemorrhage (SAH) is a common stroke subtype and is often caused by aneurysmal rupture. Accurately predicting aneurysmal SAH (aSAH) clinical outcomes may facilitate timely and effective treatment. METHODS: A dataset encompassing 12,070 aSAH Australian patients was analysed using multivariate logistic regression and machine learning techniques. Models were designed to predict the need for a surgical procedure, intensive care unit (ICU) admission, the need for continuous ventilation, prolonged length of stay (LOS), mortality, and discharge status. RESULTS: Surgical intervention was positively associated with younger age and female gender and negatively associated with indigenous Australian status. ICU admission was positively associated with age, female gender, surgery and indigenous status. The need for continuous ventilation was associated with age and ICU admission and negatively associated with female gender and surgery. Prolonged LOS was associated with older age, undergoing surgery and requiring ICU admission. Mortality was negatively associated with younger age, female gender and receiving surgical treatment. The Monash SAH Scores were developed, consisting of four scales that may be used to efficiently facilitate risk stratification and decision making in clinical practice. Machine learning models were capable of accurately predicting mortality with strong performance. CONCLUSIONS: This nation-wide study identified risk factors associated with acute clinical outcomes amongst aSAH patients. A simple scoring system was developed that may be readily applied to clinical practice to facilitate aSAH patient risk stratification and treatment planning.