This interview is part of our new AI in Healthcare series, where we interview the world's top thought leaders on the front lines of the intersections between AI and healthcare.
In this interview, we speak with Dr. Michael Abramoff, founder and CEO of IDx, to understand how his company is using AI to transform healthcare, and what the future of the industry holds.
1. What’s the story behind IDx? Why and how did you begin?
MA: Over 20 years ago, I was training as an ophthalmologist. I noticed how much time I was spending assessing people for diabetic retinopathy, a leading cause of blindness. Most of the patients I saw did not have the disease, while people who were going blind had to wait months to be diagnosed.
I realized there might be better ways, and given my experience in machine learning and software development, began to study how to automate the process of diagnosing diabetic retinopathy so that patients could be tested more efficiently and disease could be caught earlier. In 2010, I founded IDx to further develop my scientific research into an autonomous AI system that could reach patients safely and effectively. IDx’s mission is to transform the affordability, quality, and accessibility of healthcare through automation.
On April 11, 2018, IDx made national news when it became the first company to receive FDA clearance for an autonomous AI system that does not require a physician to interpret the result. An autonomous AI system does not require a physician, or in fact any human, to make the clinical diagnosis. The system, called IDx-DR, was cleared after rigorous validation to detect diabetic retinopathy, a complication of diabetes that places 30 million Americans at high risk of vision loss. But this clearance decision has wider implications beyond diabetes and eye care; it marks a huge step forward for autonomous AI solutions across all specialties and promises to transform healthcare delivery as we know it.
Another historic event occurred in May 2019, when the American Medical Association’s (AMA) CPT Editorial Committee created a new CPT code for autonomous AI — specifically for automated point-of-care retinal imaging — which facilitates reimbursement of IDx-DR for the physicians who are currently using the system.
2. Please describe your use case and how IDx uses AI:
MA: IDx-DR is an autonomous point-of-care diagnostic system that uses AI to enable non-eye care providers to detect diabetic retinopathy in primary care and retail clinics, in real-time, and at the point-of-care. Any frontline staff member can operate the AI system, which helps take high-quality images and then diagnoses them. Early, frontline detection of this disease is critical because most patients do not experience symptoms until it's too late. In fact, studies have shown that only 15 percent of Medicare-insured people with diabetes get this required eye exam regularly. Healthcare administrators, physicians, and payers have been trying to close this care gap unsuccessfully for years.
IDx-DR promises to close this care gap by removing barriers to patients receiving this test. First, it is completed at the point-of-care, where these patients have their diabetes managed. Two, it is immediate so the diagnosis can be included in the patient’s diabetes management plan, and do not need to be called back days later to hear the result — whenever this integrated management is impossible.
IDx-DR was the first-ever autonomous AI diagnostic system to receive FDA clearance, in any field of medicine. To better understand why IDx-DR’s clearance is so significant, it’s important to understand the difference between autonomous and assistive AI in the context of healthcare. Assistive AI can help physicians make decisions; for example, it might flag specific areas in an image for a physician to look at, but the physician is still responsible for the clinical decision. In contrast, an autonomous AI system makes a clinical decision without a physician being involved in that decision.
This requires rigorous validation of the safety, efficacy, and equity of the autonomous AI, and it implies that the developer of the autonomous AI assumes liability for this clinical decision. These principles are now part of the AMA’s recommended policies on augmented intelligence.
Why is it so important to distinguish between autonomous and assistive AI? Because for years healthcare costs have skyrocketed as healthcare productivity has declined. Assistive applications can improve quality of care, but they do not eliminate the need for the time and expertise of an expensive specialist physician, and thus do not address the productivity issue.
Thus, the shift from assistive to autonomous AI has the potential to increase healthcare productivity and thereby lower costs by freeing physicians from routine tasks. And, because it enables frontline healthcare providers to provide specialty diagnoses at the point of care, it can increase patient access to early disease detection.
3. Could you share a specific customer/user that benefits from what you offer? What has your service done for them?
MA: Here is one customer’s story that has been shared publicly.
IDx-DR has enabled this diabetes clinic to diagnose their patients in real time during their visit, which eliminated a four-month appointment wait time for a diabetic retinopathy exam. Now, the clinic is able to provide same-day referrals for those patients who are diagnosed with diabetic retinopathy and have a high risk of vision loss.
Other health systems have reported that they have been able to test and identify patients with the disease who need to see an ophthalmologist that would have previously been missed.
4. What other AI use cases in healthcare are you excited about?
MA: I am most excited by autonomous AI for both diagnosis and treatment, in the eye and also in other organs such as the ear, and providing patients with greater access to high-quality, safe, and efficacious diagnosis and treatment that is currently only available from specialists in a limited number of locations.
5. Where will your company be in 5 years?
MA: We are expanding rapidly into primary care and retail clinics, introducing point-of-care, real-time, autonomous AI where the patients are. Right now, we are helping people with diabetes with IDx-DR, but we are also in discussions with the FDA about the validation of similar autonomous AI for macular degeneration, glaucoma, as well as the treatment of ear infection. All of these products are in line with our mission to reduce costs, increase access and improve quality of health care and improve patients’ lives.