Artificial intelligence to enhance prehospital stroke diagnosis and triage: a perspective.

As health systems organize to deliver the highest quality stroke care to their patients, there is increasing emphasis being placed on prehospital stroke recognition, accurate diagnosis, and efficient triage to improve outcomes after stroke. Emergency medical services (EMS) personnel currently rely heavily on dispatch accuracy, stroke screening tools, bypass protocols and prehospital notification to care for patients with suspected stroke, but novel tools including mobile stroke units and telemedicine-enabled ambulances are already changing the landscape of prehospital stroke care. Herein, the authors provide our perspective on the current state of prehospital stroke diagnosis and triage including several of these emerging trends. Then, we provide commentary to highlight potential artificial intelligence (AI) applications to improve stroke detection, improve accurate and timely dispatch, enhance EMS training and performance, and develop novel stroke diagnostic tools for prehospital use.

Poor prediction of stroke mimics using validated stroke mimic scales in a large academic telestroke network.

INTRODUCTION: Telestroke enables timely and remote evaluation of patients with acute stroke syndromes. However, stroke mimics represent more than 30% of this population. Given the resources required for the management of suspected acute ischemic stroke, several scales have been developed to help identify stroke mimics. Our objective was to externally validate four mimic scales (Khan Score (KS), TeleStroke Mimic Score (TS), simplified FABS (sFABS), and FABS) in a large, academic telestroke network. METHODS: This is a retrospective, Institutional Review Board-exempt study of all patients who presented with suspected acute stroke syndromes and underwent video evaluation between 2019 and 2020 at a large academic telestroke network. Detailed chart review was conducted to extract both the variables needed to apply the mimic scales, the final diagnosis confirmed by final imaging, and discharge diagnosis (cerebral ischemic vs stroke mimic). Overall score performance was assessed by calculating the area under curve (AUC). Youden cutpoint was established for each scale and used to calculate sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and accuracy. RESULTS: A total of 1043 patients were included in the final analysis. Final diagnosis of cerebral ischemia was made in 63.5% of all patients, and stroke mimic was diagnosed in 381 patients (36.5%). To predict stroke mimic, TS had the highest AUC (68.3), sensitivity (99.2%), and NPV (77.3%); KS had the highest accuracy (67.5%); FABS had the highest specificity (55.1%), and PPV (72.5%). CONCLUSIONS: While each scale offers unique strengths, none was able to identify stroke mimics effectively enough to confidently apply in clinical practice. There remains a need for significant clinical judgment to determine the likelihood of stroke mimic at presentation.