What is the need you set out to address?

Chris: I was very aware from my work as a neurosurgeon, and from my time with Biodesign, that developing a new device or drug to treat a patient wasn't enough. Challenges with workflow in healthcare would likely mean that not all patients benefit from available treatments. There had to be a better way to make healthcare work faster and smarter.

We started in ischemic stroke, where time really matters. Our initial focus was on thrombectomy, which is a very effective treatment for a large vessel occlusion stroke. But only about two percent of patients actually get the treatment, and the reason came down to luck. If the patient was taken to a comprehensive stroke center like Stanford, a radiologist, a neurologist, and a neurosurgeon would all be on-hand to coordinate care and get the patient into treatment. But if the patient was at a different hospital, they would have to depend on an emergency room physician, who rarely sees a large vessel occlusion stroke and may not be familiar with thrombectomy treatment. So, the chances of being quickly transferred and effectively treated were really low. Back when Viz.ai were starting out, the median time to treatment for stroke patients was three hours for a comprehensive center, but five hours for other facilities without a specialist treating physician.  

In the stroke world, we say ‘time is brain’ because every minute two million patient brain cells die. Stroke is the number one cause of long-term disability in the United States. So, bringing the time to treatment down from five hours to under an hour delivers huge benefits to patients and their families, and it substantially reduces the burden on the health system and payers.  

What key insight was most important to guiding the design of your solution?

We knew that an AI-based technology could interpret a scan, but we realized that to actually make it clinically useful, we needed the interpretation to drive workflow. So, our focus wasn't about helping radiologists be better at their jobs, it was about coordinating care between the emergency room, radiology, neurology, and neurosurgery. And that insight has translated into everything we do, whether it's a condition where minutes matter, like stroke, or taking diagnosis from 5 years down to five days, like hypertrophic cardiomyopathy. I believe that the medical innovation field will find ways to cure almost every disease through the amazing work happening in biotechnology and medical devices, but the problem is that so few patients are going to benefit from that unless we have a system that directly connects the patient who needs the treatment with the person who can diagnose and prescribe or perform the treatment.

Another big insight was centered on democratizing care. We realized that AI-driven technology, that worked in the same manner every time could help by reducing the bias that we have in human healthcare. This bias might be based on which hospital the patient attends, what time of day or night it is, how tired the doctor is, and what cases the particular doctor is familiar with. From our very first FDA study in stroke, we were targeting a reduction in time-to-treatment, and we accomplished that by showing an hour-long time decrease with the software. But if you look at the study results, we also reduced the standard deviation in time-to-treatment by two hours. We took it from 127 minutes down to seven minutes. That’s what our technology can do very well – reduce variability of care. It works the same at any time of day and whether the patient is at Stanford or in rural Mississippi. And it's not reliant on the clinical literature that each doctor has read or the conferences they've recently attended. Hence, we’re able to truly democratize care.

"That’s what our technology can do very well – reduce variability of care. It works the same at any time of day and whether the patient is at Stanford or in rural Mississippi."

How does your solution work?

We like to contrast our technology with an electronic health record. The electronic health record is the system of record in a hospital, storing an exponentially increasing amount of medical data and enabling note taking, billing, etc. In contrast, Viz.ai is a system of action. It transforms workflows by facilitating the fast and accurate detection of diseases, using an AI read of imaging and other patient data, and automatically sharing relevant scans and other information to enable quicker transfers and better collaboration between the care teams at treating and referral hospitals to accelerate time-to-treatment.