Practical Guide to Implementing AI in Telestroke Networks

Stroke systems of care have been developed over many decades. Much of this process refinement has been driven by the need to evaluate for thrombolytic use in acute stroke, or by the ever-changing landscape of thrombectomy. With the development of reliable artificial intelligence (AI) algorithms, the use of AI in stroke pathways has become widely utilized. Systems of care have pivoted to incorporate this technology to optimize the systems of stroke care to improve patient outcome. Telestroke networks have exponentially increased over the past two decades and have become a standard process within the overall stroke care pathway. Various uses of AI are incorporated into many telestroke networks, though assessing how to implement these technologies into the telestroke landscape, and operational guidelines for their incorporation, have been limited. In this article, we aim to discuss the use of AI in telestroke workflows, specifically with emphasis on different platforms, technology and infrastructure requirements, operational/clinical workflow considerations, and use within transfer processes. Our aim is to provide a practical roadmap with the AI tools currently used in telestroke, and detail the critical factors which may come into play during implementation, deployment, and clinical use of this rapidly expanding tool in telestroke.