Simple-to-use emergency treatments for stroke are unavailable in rural and poor areas of the world because there are not enough physicians to evaluate and diagnose the patients. To solve this problem, NeuroSpring is developing an artificial intelligence diagnostic (AID) known as the Faucet™ that can diagnose stroke patients and make emergency treatment decisions for them without the routine need of stroke-specialized physicians. With a ubiquitously-available AID like the Faucet, we will be able to increase the number of stroke patients receiving the standard-of-care emergency treatment – intravenous tissue plasminogen activator (rtPA) – from an estimated 200,000 to 6.4 M worldwide. Increasing the availability of emergency stroke treatment with an AID would save an estimated 2.4 M quality-adjusted-life-years and $155 B USD in healthcare costs and lost productivity each year. Indeed, this estimate is likely low given the more severe impact stroke has in rural and poor areas of the world, nor does it consider the advent of new emergency stroke treatments that will have greater safety, efficacy, and availability than rtPA.
The current clinical diagnostic process for stroke is well-established and has already been mastered by the Mirtskhulava AID, which now needs to be enabled with natural language capabilities, cloud-based availability, and visual sensor capabilities to be clinically useful in the emergency diagnosis of stroke patients. Indeed, the medical and technical challenges faced by the Faucet program appear relatively straightforward. Instead, the major challenges we face are (1) ensuring the Faucet’s diagnostic accuracy, (2) securing a regulatory pathway for Faucet approval, (3) obtaining reimbursement for the Faucet service, and (4) driving market adoption. We address the Faucet program’s major challenges here, and discuss the rationale for why the emergency treatment for stroke provides a unique means for AI diagnostics to enter medical practice on a level equivalent to – not merely supportive of – the physician.