By Thomas Meinel

Twitter: @TotoMynell

Artificial Intelligence and Digital Health Technologies

The session on “Artificial Intelligence and Digital Health Technologies in Stroke” convened with an engaging array of presentations highlighting the latest advancements in stroke detection, management, and rehabilitation.

The session commenced with Teresa Ullberg from Malmö, Sweden, shedding light on wearable technology’s role in stroke detection. Those include accelerometers, EEG and carotid artery monitoring that are geared towards high-risk patients in the community. Although not clinically applicable yet, Ullberg’s presentation underscored the potential of wearable devices, emphasizing their significance in enabling prompt intervention. Given their potential for creating false alarms, such devices need to be tested rigorously in clinical trials. 

Then, Jonathan Wenstrup from Copenhagen, Denmark gave his experience with AI-based call handler recognition. It is currently difficult for call handler to correctly identify stroke calls. An AI-based model was able to improve correct classification of emergency calls without increasing false-positive dispatches. Based on those promising retrospective and simulation results, a randomized controlled trial is in the planning to understand the impact if such an AI would be implemented. 

Following on, Greg Albers from Stanford, United States discussed clinically available AI platforms in acute stroke and their impact on stroke pathways. Those platforms have been shown to reduce unnecessary transfers, speed door to puncture times, detect residual perfusion deficits, identify LVOs on NCCT and quickly identify and Quantify ICHs. A particular field of interest is the development of post-thrombectomy cone-beam perfusion acquisition that can detect residual distal occlusions and hypoperfusion as recently shown in the AFTERMATH study. However, caveats in the use of those novel techniques include that physician oversite is mandatory, the burden of false-positive alarms and the clinical impact of its use is rarely tested. 

Continuing the discourse on technological innovations, Friedhelm Hummel from Geneva, Switzerland, delved into technologies for evaluating prognosis estimation and technology based rehabilitation. This includes an individualized prognosis assessment based on connectomics and tailored, combined treatments (EEG-based TMS). Hummel elucidated on cutting-edge technologies facilitating comprehensive evaluation of neurological impairment and personalized rehabilitation strategies. 

Overall, the session provided a comprehensive overview of the evolving landscape of artificial intelligence and digital health technologies in stroke research, showcasing their transformative potential in enhancing stroke detection, management, and rehabilitation paradigms. Physicians can look out for tools that could complement or help in their setting and engage with industry partners to leverage the potential.