Artificial intelligence (AI) and the role it can play in improving acute ischaemic stroke care protocols took centre stage at the Barts Research and Advanced Interventional Neuroradiology conference (BRAIN 2021; 13–16 December, London, UK). Ameer Hassan, head of the Neuroscience Department, and director of Endovascular Surgical Neuroradiology and Clinical Neuroscience Research, at the Valley Baptist Neuroscience Institute in Harlingen, USA, reported benefits relating to patient outcomes as well as time and cost savings following the introduction of a novel, AI-powered smartphone application at his centre.
Hassan, who is also president of the Society of Vascular and Interventional Neurology (SVIN), started his presentation by noting that the current utilisation of mechanical thrombectomy is “much lower than it should be” in the USA. He cited recent data that indicate as many as 20% of acute ischaemic strokes are caused by a large vessel occlusion (LVO)—and yet rates of thrombectomies, which are widely considered to be the gold standard in LVO stroke treatment, range from 8–9% to as low as 2% in some US regions.
Discussing the various patient volume-related challenges that this may lead to, Hassan highlighted lengthy treatment times in transferred patients, hospital bed shortages, healthcare provider burnout, and increased costs due to delays in delivering thrombectomies. After asserting the importance of treating more patients more quickly, he went on to state that AI solutions “really do make a difference” on this front.
One way of implementing these solutions is during the detection phase prior to a patient being transferred to the cath lab for a thrombectomy procedure, Hassan told the BRAIN conference audience. “If we can detect [a stroke] earlier, rapidly triage, synchronise our care coordination, and then transfer the patient, we could get them treated much more quickly,” he said.
Hassan went on to compare the standard stroke workflow in many referral centres—a complex process that can take more than two hours, and involves multiple stages and specialists—with a synchronised stroke care model using the Viz intelligent care coordination platform (Viz.ai). The latter process, he said, deploys AI analysis, and uses a smartphone application to alert multiple specialists of the LVO diagnosis within minutes, potentially cutting the detection timeline down to as little as 39 minutes.
Elaborating on these claims, Hassan presented a report, entitled “Early experience utilizing artificial intelligence shows significant reduction in transfer times and length of stay in a hub and spoke model”, that has been published in Interventional Neuroradiology (INR). The report recounts a multicentre, retrospective study evaluating the impact of Viz on the time between computed tomography (CT) angiogram acquisition at a primary stroke centre and arrival at a comprehensive stroke centre among 43 LVO stroke patients.
Comparing these timelines at the Valley Baptist Neuroscience Institute over two time periods—pre-Viz implementation (February 2017 to November 2018) and post-Viz implementation (November 2018 to May 2019)—researchers observed a 66-minute time saving, on average (171 vs 105 minutes; 39% reduction), which also resulted in a 55% reduction in neuro-intensive care unit (ICU) length of stay (3.5 days; 6.4 vs 2.9 days), for patients who underwent a mechanical thrombectomy.
“Patient outcomes are time-dependent—we already know this,” Hassan continued. “Every minute saved leads to more disability-free life. So, if we say that each minute of delay in receiving a thrombectomy results in a loss of four days of disability-free life, and 10 days of functional independence [0–2 modified Rankin Scale score], 66 minutes could mean nearly 265 days of additional disability-free life and 660 days of additional functional independence.”
Quipping that “time is money—not just brain”, Hassan noted that those 66 minutes could also be worth just under US$70,000 in cost savings as well, due to each minute-long delay prior to receiving a thrombectomy procedure holding an economic value of US$1,059, on average. Discussing the 2.5-day reduction in overall length of stay observed in the INR study, he also told the audience that this reduction could be worth roughly US$12,300 to the hospital itself as well.
In addition to touching on data from other centres that have deployed Viz in their stroke care protocols, including significant time savings and improved clinical outcomes at Mount Sinai Health System (New York, USA), and a door-to-device time of less than 90 minutes in all thrombectomy cases at Sky Ridge Medical Center (Lone Tree, USA), Hassan concluded his presentation by stating that this novel technology “directly correlates to better outcomes—including increased access to care for stroke patients”.