AI in Acute Stroke Care: Current Evidence, Challenges and Future Directions

Background: Acute stroke requires rapid and accurate clinical decision-making, and artificial intelligence (AI) is increasingly being integrated into diagnostic and treatment workflows.  

Methods: A narrative review of peer-reviewed literature and relevant professional and public-health guidance was undertaken, focusing on AI applications in neuroimaging, large-vessel-occlusion detection, treatment selection, prognosis, triage, and workflow optimization.  

Results: AI can facilitate rapid analysis of non-contrast CT, CT angiography, and perfusion imaging; detection of intracranial hemorrhage and large-vessel occlusion; infarct-core estimation; outcome prediction; and automated communication of time-critical findings. These applications may improve diagnostic consistency and workflow efficiency. However, challenges include dataset bias, limited generalizability, explainability, interoperability, cybersecurity, automation bias, and limited prospective validation.  

Conclusion: AI has considerable potential to enhance acute stroke care but should complement rather than replace clinical expertise. Future research should emphasize multicenter validation, equity, usability, cost-effectiveness, and patient-centered outcomes.