The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project is to arm emergency personnel with an objective tool for identifying large vessel occlusion (LVO) strokes while in route to the hospital. This rapid and accurate triage of stroke will enable routing of patients to the most appropriate care setting and reduce the time to intervention. LVOs require endovascular therapy which only comprehensive stroke centers have the capability to conduct. If a patient with an LVO is routed to a hospital without endovascular capabilities simply because it was closer, the time to intervention is extended drastically. When it comes to improving outcomes, time to optimal intervention is the most important factor with the best outcomes achieved under three hours and statistically significant improvements for each 15-minute window under that threshold. Stroke is the second leading cause of death and the primary cause of long-term disability worldwide costing the US $65B every year. Nearly 800,000 people suffer a stroke in the US annually and 40% are left with a permanent disability. The project will streamline stroke triage in the pre-hospital setting to reduce time to intervention and improve outcomes in stroke patients. This Small Business Innovation Research (SBIR) Phase I project an EEG-based product for EMS workers to use in the pre-hospital setting for the fast and objective diagnosis of LVO in suspected stroke patients. In under five minutes, EMS workers will be able to deploy, collect data, and have the analyzed results presented in an intuitive dashboard identifying the probability of an LVO, enabling EMS workers to route patients to stroke centers with EVT capabilities. When a patient arrives at the hospital, the determination made within the ambulance will be conveyed to physicians who can then immediately start intervention, reducing the time from onset to intervention and improving short and long-term patient outcomes. Comprehensive historical datasets of EEG-data from stroke patients using a broad array of hardware will be used to develop a machine learning model that can classify patients into LVO vs non-LVO stroke and stroke vs non-stroke. Automation of data cleaning and feature extraction will enable a highly user-friendly experience and the required workflow integration for our end-users, emergency medical technicians. Lastly, this model will be validated with novel EEG data collected at two clinical sites, laying the foundation for regulatory interactions.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
