Dr Google wants to become your optician

Two years after unveiling connected contact lenses measuring glucose levels in the blood, the web giant reaffirms its involvement in eye care with a press release published in the
American Medical Association newspaper announcing the creation of An algorithm using photos from the back of the eye (method applied by oculists) to identify the symptoms of diabetic retinopathy.

To achieve this, the Mountain View firm collaborated with a team of specialists from the United States and India. The group created a database of 128,000 images to form a deep neural network (wired computers like the brain).


It goes without saying that much work will be needed before the algorithm can be used by the general public, but Google is optimistic: "The algorithm could encourage patients to consult doctors earlier in the future."

In a blog post, Dr. Lily Peng at Google Research adds, "Automated and highly accurate screening methods have the potential to help general practitioners diagnose more patients and quickly refer them to a specialist when it is necessary. We hope that this study will be the beginning of a long series of examples demonstrating the ability of automated science to solve important problems and optimize the quality of health care. "

As a reminder, diabetic retinopathy alters the blood vessels of the retina, frequently causing a decrease in vision or even blindness in people with diabetes.