Paul Karpecki, OD, FAAO, chief clinical editor at Review of Optometry, quickly guided the audience into Session Four: “A-Eye: How Artificial Intelligence is Transforming Eyecare,” immediately introducing ways in which artificial intelligence can directly enhance the ophthalmic profession.
Karpecki referenced a project he has been working on for several years, Ophthalmic Resources, in which patient information is collected in an EMR and an algorithm is used to help determine the best treatment for patients with certain conditions, like dry eye. In this example, the algorithm combed through patient feedback to determine how well they fared from teardrops they had been prescribed. The stunning result: “With AI, we found we had been wrong 90 percent of the time,” Karpecki said.
Fast, Safe and Easy Diagnosis
AI can work in tandem with eyecare practitioners, and even outperform them. This was evidenced in a video interview conducted by VM’s Andrew Karp with Pearse A. Keane, MD, FRCOphth, consultant ophthalmologist at Moorfields Eye Hospital in London.
In the prerecorded interview, Keane explained how Google’s DeepMind, a leader in artificial intelligence research and its applications in health care, can analyze OCT scans. It does this by using deep learning, a branch of machine learning that deals with building and training neural networks with multiple layers, most often excelling at tasks that involve sorting and classifying data.
When Keane approached DeepMind co-founder, Mustafa Suleyman about applying deep learning techniques to disease diagnosis, they found that the technology could change the way doctors treat and diagnose retinal disease, and that it was exceptionally good at identifying and predicting eye diseases like macular degeneration and diabetic retinopathy.
“Any retinal disease you can diagnose by looking at a retinal OCT scan, we think we can diagnose using AI diagnosis,” said Keane, who is spearheading a collaboration between DeepMind and Moorfields, which performs thousands of OCT scans each month.
He is especially optimistic about applying AI to OCT scans because they are noninvasive and safe, and can be administered several times without putting a patient’s eye health at risk. “It takes 30 seconds, whether you’re a five-year-old or a 95-year-old,” Keane said. “There is a very unique opportunity with OCT imaging in this regard. Patients can have multiple scans done over time without causing damage like X-Rays do.”
And, he added, this not only makes treatment more available to “a huge number of patients,” it also creates longitudinal data and long term follow ups. “Indeed, our patients who are receiving treatment for AMD will have OCT scans every month or so,” he said.
Keane’s second focus for the time being is diabetic eye disease, which he described as a “public crisis.” He explained that the U.K. has a national screening program for diabetic retinopathy, but that the retinal photos taken in these cases are not as high resolution as OCT scans. AI would make administering OCT scans in national screening programs financially possible, he said.
Overall, Keane expressed a confident outlook regarding AI’s applications in the eyecare field. “We can image almost every part of the eye with cellular level resolution. Ophthalmology will be the first branch of medicine to be fundamentally transformed by AI,” he concluded.
AI and Global Eyecare Policy
| || |
| || Review of Optometry’s Paul Karpecki, OD, FAAO, outlined a project called Ophthalmic Resources, in which patient information is collected in an EMR and an algorithm is used to help determine the best treatment. |
| || |
| || |
| || Pearse A. Keane, MD, FRCOphth, consultant ophthalmologist at Moorfi elds Eye Hospital in London (above) appeared in a video interview with VM’s Andrew Karp. Keane discussed how Moorfi elds’ collaboration with Google’s DeepMind, a leader in AI research, is helping to advance vision care. |
| || |
| || |
| || Kovin Naidoo, OD, PhD, FAAO, CEO, Brien Holden Vision Institute spoke about using data to manage eye conditions around the world. |
| || |
| || |
| || Cool Doctors’ John Gelles, OD, FIAO, FCLSA, emphasized that using telemedicine like Eyecarelive enables a “hybrid office,” combining physical and virtual care. |
| || |
After the video, Kovin Naidoo, OD, PhD, FAAO, CEO, Brien Holden Vision Institute, spoke about using data to manage eye conditions around the world, particularly in relation to the Global Burden of Disease (GBD). “Increasingly, practitioners and policymakers are finding value in GBD projections and graphs that are available … which can be used to affect policy change and advocacy efforts,” Naidoo said.
He also referenced a “public health crisis”—myopia. “By 2050, half of the world will be myopic,” Naidoo said, proposing that eye doctors must look for and embrace the “innovative, cost effective solutions to obtain data” that AI offers.
Naidoo referenced two technologies that he believes can help alleviate global eyecare crises: Portable Eye Examination Kit (PEEK) and Microsoft Intelligent Network for Eyecare (MINE). The latter, launched by Microsoft India in collaboration with L.V. Prasad Eye Institute last year, operates with the global mission to build an AI platform for eyecare.
“Historically, gathering data has been expensive, labor intensive and costly,” Naidoo said. “We’re not going to make effective change unless we effect policy change. Having shared data will allow us at a national level to influence policy in various geographical areas.”
Embracing the ‘Hybrid Practice’
Following Naidoo was John Gelles, OD, FIAO, FCLSA chief of emerging technologies at Cool Doctors, a telemedicine platform used by optometrists and ophthalmologists. He introduced Eyecarelive, a cloud-based patient-doctor platform designed for eyecare professionals to see their patients online between in-person visits.
“We all practice some form of telemedicine,” Gelles said during a panel following his presentation. Calling in prescriptions and communicating about patients with other practitioners are just two examples of this, he said. “Now we can do this with graphs and big chunks of data that can improve the care of the person in your chair.”
Gelles emphasized that using telemedicine like Eyecarelive enables a “hybrid office,” combining physical and virtual care. This can increase efficiency and improve doctor-patient communication by allowing more accurate, timely data. “What we’re trying to do is facilitate care of the future ... by extending our reach, we can create better outcomes,” he said.
Ocular conditions he believes are suitable for telecare include glaucoma and diabetic retinopathy, which require self-administered patient maintenance such as IOP and glucose level measurements.
“A program like this is a massive time saver for the doctor and the patient. Information is going directly from the patient to meaningful statistics right away, saving time, money and resources. In a traditional scenario, a person has to come to my clinic four times in a day. The dropout rate is huge when we have to do that.”
Utilizing this form of artificial intelligence, Gelles said, would introduce more consistency, accuracy and repeatability to the treatment and management of certain eye diseases. “AI will back up the doctors, not replace them. It’s making the doctors better doctors.”