A study in JAMA found that a Google-developed automated deep learning algorithm is able to interpret signs of diabetic retinopathy, an eye disease that sometimes results in blindness. The study was led by Lily Peng, MD, PhD, who works at Google Research.
The researchers used two validation sets of 9,963 images and 1,748 images to compare the performance of the automated deep learning algorithm with the performance of manual grading by ophthalmologists when identifying diabetic retinopathy from retinal fundus photographs.
When detecting referable diabetic retinopathy, the algorithm had high sensitivities (97.5 percent and 96 percent) and high specificities (93 percent and 94 percent) in both validation sets. The researchers concluded that "in this evaluation of retinal fundus photographs from adults with diabetes, an algorithm based on deep machine learning had high sensitivity and specificity for detecting referable diabetic retinopathy."