The predictive performance of artificial intelligence on the outcome of stroke: a systematic review and meta-analysis

Objectives: This study aimed to assess the accuracy of artificial intelligence (AI) models in predicting the prognosis of stroke. Methods: PubMed, Embase, and Web of Science were searched from inception to February 2023 for cohort studies using AI to predict acute stroke prognosis. Study quality was assessed with QUADAS, and a random/fixed-effects approach was used to summarize data. The area under the receiver operating characteristic curve (AUC) was used as the accuracy metric. Results: Of 1,241 publications, seven studies were included (17 AI models). Risk of bias was low and heterogeneity was not significant. The pooled AUC under a fixed-effects model was 0.872 (95% CI 0.862–0.881). Subgroup AUCs: deep learning (DL) 0.888 (95% CI 0.872–0.904), logistic regression (LR) 0.852 (95% CI 0.835–0.869), random forest (RF) 0.863 (95% CI 0.845–0.882), support vector machine (SVM) 0.905 (95% CI 0.857–0.952), and XGBoost 0.905 (95% CI 0.805–1.000). Conclusion: AI models demonstrate good accuracy for predicting outcomes in ischemic stroke and can assist clinicians. Continued algorithmic advances and larger datasets may further improve performance.