Email
Password
Remember meForgot password?
    Log in with Twitter

article imageQ&A: New AI system that tests for diabetic eye disease Special

By Tim Sandle     Jul 6, 2018 in Health
Des Moines - The U.S. now officially have access to an artificial intelligence system that tests for diabetic eye disease in primary care. The system does not require patients to see an eye care specialist. Digital Journal spoke with the people behind the platform.
The early detection of diabetic eye complications (diabetic retinopathy) is an important component of comprehensive diabetes care. The new autonomous diagnostic artificial intelligence platform represents an important advance in the field of early diagnosis.
The new system is called IDx-DR and it has been used in patient care in the European Union for several years. However, the system has just received FDA clearance for use in the U.S. The platform is now in use in patient care at the University of Iowa Health Care. To discover more, Digital Journal discussed the technology with UI Health Care administrators and the developers behind IDx-DR: Maia Hightower, MD, MBA, MPH, chief medical information officer for University of Iowa Health Care; Yumi Imai, MD, University of Iowa associate professor of internal medicine – endocrinology and metabolism; Ryan Amelon, PhD, Director of Research and Development, IDx; and Michael Abramoff, MD, PhD, President and Founder, IDx.
Digital Journal: What impact is technology having on healthcare?
Maia Hightower: Technology is having a tremendous impact on healthcare. In general, there are currently three main forces driving change in the healthcare industry: payment reform, the rising role of the consumer, and technology. As for technology, digital imaging has fueled AI development for innovations like IDx-DR, creating new solutions to the problems we have in healthcare. Arguably advances in health technology are underlying the other two forces, payment reform and the rising role of the consumer. The technology experience of consumers outside of healthcare raises the bar of expectation for our consumers when they interact with the healthcare system.
DJ: How important is artificial intelligence becoming for healthcare?
Hightower: AI is a result of digitization of the healthcare industry including digital imaging and myriad of other health IT solutions. Artificial intelligence may provide us with the innovative breakthrough solutions that allow us to solve some of our biggest problems in healthcare.
DJ: What the general advantages?
Hightower: The general advantages of AI include automation of certain tasks. This automation allows for increased scale, i.e. increased access, to a service. For systems like IDx-DR, it also allows talent to shift focus to other high priority areas. Healthcare is an industry where there is a critical shortage of key talent including medical assistants, nurses, and physicians. AI helps to detect both operational and clinical high-risk areas so that limited resources can be targeted to areas of highest need or greatest return.
DJ: What is diabetic eye disease and how common is it?
Yumi Imai: Diabetic retinopathy (DR) is an eye disease that is associated with high blood glucose level in diabetes. DR is a leading cause of vision-loss globally. Of an estimated 285 million people with diabetes mellitus worldwide, approximately one third have signs of DR, which is what IDx-DR tests for. Early detection of DR is very important as treatment can significantly reduce the risk of vision-loss.
DJ: How did you develop the AI system that tests for diabetic eye disease?
Ryan Amelon: IDx was founded by clinician scientists who were heavily involved in the AI development process and spent years researching use of autonomous AI for medical diagnosis. After founding IDx, they worked closely with the R&D team, breaking down the diagnostic process into a series of steps, essentially outlining the framework that the AI system would follow. AI components were then built to replicate each of these steps, capturing a similar process that a retina specialist would use for diagnosis using the same images.
DJ: How does the technology work?
Amelon: Due to the highly robotic nature of the camera and the fully-autonomous diagnosis, virtually anyone in a healthcare setting can be trained to operate IDx-DR. A person with diabetes has two images of the back of the eye taken with an automated fundus camera, which are then processed by IDx-DR using a laptop next to the camera. IDx-DR automatically assesses imaging protocol, the quality of the images, and the presence of disease. The result is displayed to the user in under a minute or inserted directly into the EMR. The entire patient experience is roughly 5 minutes.
DJ: How has the use of the system gone in Europe?
Michael Abramoff: IDx-DR is rapidly gaining adoption across Europe and is currently in use in the Netherlands, Austria, and Spain. Additional pilots are scheduled in Italy, Czech Republic, and Finland.
DJ: Is any clinical input required?
Abramoff: I’ve worked on developing the science behind diagnostic AI for decades, and eventually I realized that in order to truly impact healthcare productivity, an autonomous use case – where physician oversight is not required - is essential. That led to the founding of IDx. Thus, IDx-DR was designed to produce a result without input from an ophthalmologist – it is a fully autonomous AI diagnostic system. In a press release, FDA also stated that IDx-DR “is the first device authorized for marketing that provides a screening decision without the need for a clinician to also interpret the image or results.”
DJ: How did you go about gaining US FDA approval?
Abramoff:Over the course of the last 8 years, IDx has had dozens of written and in person interactions with the FDA’s Division of Ophthalmic and Ear, Nose and Throat Devices. This was not an easy or short road, but the knowledge we gained through the FDA process became foundational to our company. Not only did this help shape the development of our clinical trial, it provided a framework for our development of future autonomous AI systems.
The FDA had never before approved an AI diagnostic system for an autonomous use case, so we designed our clinical trial to establish to FDA that an autonomous use case could be safe for patients. This required using an explainable framework for our algorithm development. This also required testing IDx-DR with prospective data in its intended use case scenario: primary care settings with novice operators. Lastly, IDx-DR was tested on a representative sample of the 30 million people with diabetes in the U.S.; the trial involved 900 subjects at 10 sites throughout the United States.
DJ: What happened following the trial?
Abramoff:After completing the trial, we submitted a request to FDA for IDx-DR to be included in the Breakthrough Devices Program, which expedites the review of medical devices that “provide for more effective treatment or diagnosis of life-threatening or irreversibly debilitating diseases.” We felt IDx-DR was qualified for this designation because the healthcare system desperately needs a more efficient and cost-effective way to detect diabetic retinopathy; too many patients go blind needlessly because they aren't diagnosed in time. FDA agreed, granting us expedited review shortly after. Less than three months later, we received FDA clearance, reaching our goal after years of working with FDA.
DJ: Where do you see healthcare AI technology heading over the next five years?
Abramoff:With the FDA’s clearance of IDx-DR, we now have a roadmap for the safe and responsible adoption of autonomous AI in medicine. In the long-term, the IDx-DR platform will be expanded to include other autonomous algorithms that diagnose different diseases, such as glaucoma, macular degeneration, cardiovascular disease, Alzheimer’s and more.
In order for widespread adoption of AI to happen, clinicians need to see how an algorithm can be developed into a complete system solution that fits into their own clinics. When individuals have direct experience working with these solutions, it can provide a great deal of reassurance. That’s why it is exciting to see UI Health Care leading the way with the adoption of IDx-DR; it’s the early adopters who help others feel more comfortable using AI.
Also, as a responsible AI company, we need to help clinicians and their patients trust AI. We can build that trust by conducting scientific studies, by developing AI-based systems that are aligned with clinical guidelines, and by ensuring external review of validation by experts and regulators, including FDA.
More about diabetic eye disease, Diabetes, Artificial intelligence
More news from