🔒Guilford biotech develops AI-enabled system to diagnose diseases through the eye

It’s said the eyes are the windows to your soul, but for Jonathan Rothberg and Vasiliki (Vicky) Demas, the eyes also are an observation deck for your physical health. Rothberg and Demas are chairman and CEO, respectively, of identifeye HEALTH, a medical device company that has developed AI-enabled software and a camera that can, among other uses, screen the retina for diabetic retinopathy, an eye disease related to diabetes. The privately held company, which has 40 employees divided between Guilford and a facility in Redwood City, California, recently registered and listed the device with the U.S. Food and Drug Administration (FDA), from which it is still seeking formal approval. That allowed it to begin using the AI-operated retina imaging system to screen patients in partnership with Remote Area Medical (RAM), a Tennessee-based nonprofit provider of remote medical clinics, and the Maryland-based Heart Health Foundation. The 15-pound identifeye Camera, which bears a passing resemblance to a Star Wars droid, is built to be portable, user-friendly, painless and noninvasive. While its ability to help diagnose diabetic retinopathy is a significant achievement — 10% of the U.S. population has been diagnosed with diabetes, and 30% of those people have some form of diabetic retinopathy, according to a University of Pennsylvania study — it’s just the beginning of what this device can do.

“Identifeye gave us the opportunity to invent a new branch of diagnostics,” Rothberg said. “When you think of diagnostics, you think of two major branches — radiology, which is imaging, and then clinical chemistry, where they take your blood and they analyze what’s in it. What I wanted to do was create a new branch of diagnostics that was synergistic between the two.” That was possible, Demas said, because of the eye’s anatomy. “The retina is a noninvasive window into the body, and health in general, beyond just kind of looking at ocular disease,” she said. “The retina is the one part in your body where you can directly look at blood vessels and nerves. It’s part of the central nervous system. So, there are a ton of disease biomarkers that now we can completely access noninvasively.” That means identifeye’s technology has the potential to eventually help diagnose not just eye diseases, but heart disease, Alzheimer’s and more. That was Rothberg’s goal for founding the company in 2018 when, he said, he had an idea to expand diagnostics and came up with “a silly name.”

Dimensional thinking

If Rothberg’s name sounds familiar, it’s likely because identifeye HEALTH was not his first bioscience startup. He founded his first company, CuraGen, while still a graduate student at Yale University. CuraGen worked to develop drugs that targeted specific genes. He went on to invent a method for high-speed DNA sequencing, bringing it to market with his second venture, 454 Life Sciences. Other startups include Hyperfine Research Inc., which developed a portable MRI, and Butterfly Network, which developed a pocket-sized ultrasound device. Rothberg said his idea for a new diagnostic method was born from the same motivation as his other ventures. “My origin stories are always the same. Every single company I start, I start because I want to have something there when somebody I love needs it,” he said. “And every single company I started since 2011 has been based on the fact that AI, artificial intelligence, now works.”

The idea to combine imaging with the ability to see the chemistry in the eye was not his own. For that, he thanked Google. “Google is an amazing company, but they have more ideas than they finish,” Rothberg said. “I saw a paper, it said AI in the eye was going to be a big deal. Google didn’t do anything with it.” So, he formed his company and decided to call it Tesseract, which in geometry is a four-dimensional hypercube. “I called it Tesseract because we would look at multiple dimensions within the eye,” he said. Armed with the Google paper and his new company, Rothberg was able to leverage that and his reputation to raise $90 million from investors, including San Francisco-based Foresite Capital.

Finding a CEO

While his startup was off the ground, Rothberg needed someone to oversee the day-to-day operations who understood both the technical and medical challenges. To create a new branch of diagnostics, he said, “you really need an extraordinary CEO, because it’s hard enough to lead a company with one domain, AI, or a second domain, radiology, but it’s harder yet to have a company that can span multiple domains and then be practical.” The company found Demas, who has a Ph.D. in chemical engineering and physical chemistry from the University of California, Berkeley. Demas also spent more than three years working at Google X, the division with the mission to invent “moonshot” technologies intended to make the world a better place. She was a founding member of the division’s Google Life Sciences team, which was later spun out to form Verily, where she led several teams and projects, including one involving retinal imaging. “Vicky was that person, because she had come from that Google AI world, but she also had that imaging background,” Rothberg said. “So, she had a foot planted in the two areas.” Demas said she was eager to take on the role "and figure out how to take the investment, and work on whatever would be the groundbreaking thing to develop the technology.” One of the first things Demas suggested was changing the company name, Rothberg said. “Tesseract is a good idea, but identifeye tells you what we’re going to do with the idea, and that was a critical transition,” he said. “She came in and we went from Tesseract, a company that was extremely well funded, a company that had the right technology, and now we would bring it to the market with identifeye, and that’s exactly what she’s done with the recent FDA approvals.”

Elegantly simple

The identifeye Camera and tablet feature a sleek, clean design and are simple to use. The tablet provides written and verbal instructions in either English or Spanish, though Demas said AI will eventually allow it to add many more languages. AI also allows the camera to get the highest-quality images possible from each patient, she said. AI-directed prompts guide the patient through the exam, taking the responsibility away from the operator to make sure the patient is set up properly to have their eyes scanned, Demas said. “Now AI does that, so a couple minutes of training and you can actually run this, because it’s meant to be intuitive,” she said. “I tell people we wanted to make sure that retinal imaging is as easy as measuring someone’s blood pressure.” The device verifies each scan to ensure diagnostic quality. “From a claim perspective, we are capturing high-quality images that can be interpreted by a qualified person,” Demas said. She added that, for now, images are uploaded to the cloud so they can be interpreted by a tele-retinal service. The company is generating some revenue, but this year’s focus is on building interest in the product, Rothberg said. Revenue will come from both software and hardware sales, with plans to expand into new uses over time. “We need to continue with that AI, because we want to go from diabetic diagnosis, which is one indication, to a general tool,” he said. “There’s no reason that you can’t understand Alzheimer’s by the eye, and there’s some evidence for that. There’s no reason you can’t understand cardiac diseases.” “AI is not magic,” he added. “It’s just the best method to do massive correlations.” Editor's Note: This article has been updated to correct errors made by the reporter, including to note that the camera has not yet been approved by the FDA.