Diabetes affects more than 415 million people worldwide, or 1 in every 11 adults. About 45% of diabetic patients are likely to have diabetic retinopathy. Early detection and treatment of diabetic retinopathy can help reduce preventable vision loss.
Artificial intelligence is the use of computers to simulate human intelligence. Deep learning uses huge amounts of data paired with extensive neural networks to give computers new capabilities; training the computers to do what our brains do naturally. Researchers at Stanford University put deep learning to work in developing an algorithm that allowed them to identify which diabetic patients needed to be referred to an ophthalmologist.
The algorithm was created based on more than 7,500 images from a wide range of patients. These images were used to teach the computer to identify between healthy patients and those with any stage of disease from mild to severe. The computer was accurate 94% of the time. The program can be run on a personal computer or a smartphone, making it potentially accessible worldwide.
The team awaits approval from the FDA before it can be used outside of a research setting. So for now, patients will have to continue to visit their ophthalmologist for their annual diabetic eye exam. In the future – you may be able to get your screening diabetic retinal exam using your smartphone!
Source: Ophthalmology, July 2017
Michelle Akler, MD
Akler Eye Center
Sterling Heights & Dearborn