A89 Machine learning for predicting functional outcomes in acute ischemic stroke

<h3>Introduction</h3> This is a AI model deveopment study using a nation-wide registry of 40,586 patients with acute ischemia. <h3>Aim of Study</h3> In this study, we aimed to address these gaps by utilizing the nation-wide stroke registry to develop and validate ML models for predicting stroke outcomes. <h3>Method</h3> We conducted a retrospective analysis of AIS patients from the Korean National Stroke Registry. Multivariate logistic regression and ML algorithms, including random forest (RF) and support vector machine (SVM), were used to assess the impact of clinical and demographic variables on discharge outcomes. Functional recovery was defined as a modified Rankin Scale (mRS) score of 0–2 at discharge. <h3>Results</h3> Using a comprehensive dataset of 40,586 patients, we developed three machine learning models—Random Forest (RF), Support Vector Machine (SVM), and Logistic Regression—to predict favorable functional outcomes (modified Rankin Scale ≤2) at discharge. Among these, the RF model revealed superior predictive performance, achieving an area under the curve (AUC) of 0.87, compared to the SVM and Logistic Regression each achieving an AUC of 0.80. Key predictors identified included the National Institutes of Health Stroke Scale score, age, onset-to-image time, onset-to-door time, Charlson Comorbidity Index, and administration of intravenous thrombolysis. <h3>Conclusion</h3> This study underscores the transformative potential of machine learning in stroke management, predicting and improving patient outcomes and streamlining healthcare delivery <h3>Conflict of Interest</h3> No