Roche AI Program Can Predict Progression of Diabetic Retinopathy
Harnessing the increasingly important applications of artificial intelligence, researchers from Genentech and its parent company Roche have developed what is being called the first “deep learning model” that can predict which patients with diabetic retinopathy will progress the fastest.
In a new paper published in Nature Digital Medicine, the researchers noted that their method uses images of the retina taken during a single doctor visit. The algorithm developed by the Roche team can then be used to determine the rate of progression in diabetic retinopathy. As the Genentech and Roche team notes, this type of predictive algorithm could help diabetic retinopathy patients get the individualized care they need. Also, ophthalmologists will be able to tailor treatment for their patients due to the understanding of how the patient’s condition is likely to progress in one to two years.
“Upon further development on larger and more diverse datasets, such an algorithm could enable early diagnosis and referral to a retina specialist for more frequent monitoring and even consideration of early intervention. Moreover, it could also improve patient recruitment for clinical trials targeting DR (diabetic retinopathy),” the researchers said in their article.
Diabetic retinopathy affects about one-third of adults with diabetes over the age of 40. In 2017, the researchers noted there were approximately 425 million people in the world with diabetes and the number is expected to increase to 642 million by 2040. Diabetic retinopathy (DR) is the most common microvascular complication of diabetes and can progress asymptomatically until a sudden loss of vision occurs. In the United States, approximately 7.7 million people have diabetic retinopathy, which can lead to vision loss, according to the National Eye Institute. That number is expected to climb to 14.6 million in 2050.
While diabetic retinopathy is a common issue, it often remains undetected until it progresses to an advanced vision-threatening stage, the researchers noted. Current screening methods, which are based on assessment of color fundus photographs, is not as accurate and leaves a large proportion of patients undiagnosed until it’s too late, the researchers said.
Genentech’s Zdenka Haskova, a medical director in clinical ophthalmology, said the research shows that future diabetic retinopathy progression in individual patients can “be predicted with high accuracy by deep learning models from retinal photographs obtained at one single initial visit.” Haskova added that the deep learning technology “paves the pathway for an AI tool that can inform management strategy with optimal check-up frequency and potential timely intervention to help preserve the vision of patients.”
The research is part of Genentech and Roche’s Ophthalmology Personalized Healthcare initiatives that aim to combine meaningful large-scale data and AI technology to prevent ocular conditions and preserve vision, the company said in a note to BioSpace. It is believed that the algorithm will be able to provide doctors with the necessary patient information that will allow them to focus on the most pressing needs and stratify patients based on the speed of progression.
“There is a high need for easy-to-use diagnostic or predictive tools to enable efficient allocation of healthcare resources with focus on patients who need the most attention,” Haskova added.
Genentech’s Lucentis has been approved for all forms of diabetic retinopathy. Earlier this year, Regeneron’s Eylea was approved as a treatment for diabetic retinopathy. The FDA granted Novartis’ brolucizumab priority review in mid-April, a treatment for wet age-related macular degeneration.