Using an Artificial Intelligence Application Expedited ENRICH Trial Patient Screening: A Single Center Experience

Abstract

IntroductionThe mantra “time is brain” has led to significant efforts to expedite tPA administration time and initiation of mechanical thrombectomy for the treatment of ischemic stroke. Evidence supporting surgical intervention for intracerebral hemorrhage (ICH) remains elusive. Numerous clinical trials have had negative overall results and others are still on-going. Most investigators and trial protocols have agreed that intervention should occur rapidly to decrease the risk of hematoma expansion and reduce perihematomal edema, however no trial has directly studied this question. Artificial intelligence applications have been shown to improve ischemic stroke workflow metrics, both decreasing transfer times from outside hospitals and rapidly alerting the interventional teams. We aimed to determine whether the implementation of an ICH detection algorithm that provides immediate active notification to provider cell phones would improve hemorrhagic stroke workflow at our institution.MethodsA retrospective review was performed of patients presenting between January 2018 and March 2022 who suffered a spontaneous ICH and for whom the neurosurgical service was consulted for possible surgical intervention. Stroke workflow metrics were compared pre- and post-implementation of the VizAI (Viz.ai, San Francisco, California, USA) smartphone application. Additional demographic, clinical, and radiographic information was all collected.Results188 adult patients were identified during the study period. Time between identification of ICH to neurosurgical team notification was reduced by 50 minutes after the implementation of VizAI (p<0.002). The number increases to 57 minutes when hemorrhages not identified by the ICH algorithm were excluded.DiscussionActive notification of the neurosurgical team by an artificial intelligence application significantly reduces the time from hemorrhage identification to surgical evaluation. Further studies are needed to evaluate whether this results in a clinical benefit.