Machine Learning to Predict Delayed Cerebral Ischemia and Outcomes in Subarachnoid Hemorrhage

OBJECTIVE: To determine whether machine learning (ML) algorithms can improve the prediction of delayed cerebral ischemia (DCI) and functional-outcomes after subarachnoid hemorrhage (SAH).

METHODS: ML models and standard models (SM) were trained to predict DCI and functional-outcomes with data collected within 3 days of admission. Functional-outcomes at discharge and at 3-months were quantified using the modified Rankin scale (mRS) for neurological disability (dichotomized as 'good' (mRS≤3) vs 'bad' (mRS≥4) outcomes). Concurrently, clinicians prospectively prognosticated 3-month outcomes of patients. The performance of ML, SM and clinicians are retrospectively compared.

RESULTS: DCI status, discharge, and 3-month outcomes were available for 399, 393 and 240 subjects respectively. Prospective clinician (an attending, a fellow and a nurse) prognostication of 3-month outcomes was available for 90 subjects. ML models yielded predictions with the following AUC (area under the receiver operating curve) scores: 0.75 ± 0.07 (95% CI: 0.64 to 0.84) for DCI, 0.85 ± 0.05 (95% CI: 0.75 to 0.92) for discharge outcome, and 0.89 ± 0.03 (95% CI: 0.81 to 0.94) for 3-month outcome. ML outperformed SMs, improving AUC by 0.20 (95% CI: -0.02-0.4) for DCI, by 0·07 ± 0.03 (95% CI: -0.0018-0.14) for discharge outcomes, by 0.14 (95% CI: 0.03 -0.24) for 3-month outcomes and matched physician's performance in predicting 3-month outcomes.

CONCLUSION: ML models significantly outperform SMs in predicting DCI and functional-outcomes and has the potential to improve SAH management.