Google Uses Machine Learning to Detect Diabetic Eye Disease

76549824 - retinal

Did you know that at this very moment there are more than 400 million diabetic patients who could be at risk for a serious condition that causes permanent blindness?

Diabetic retinopathy is a condition that can affect people with both type 1 and type 2 diabetes.

Fortunately, eye doctors can treat the disease if they are able to catch it early. But if it’s left untreated for too long, it can cause irreversible blindness. This is why it’s so important for diabetes patients to undergo annual eye exams.

  • During the eye exam, a doctor will analyze photos that show the back of a diabetic person’s eyes.
  • These photos can be used to detect the presence of diabetic eye disease, and determine how advanced the disease is in people with positive diagnoses.

Annual screenings are essential because they allow specialists to catch diabetic retinopathy early and prescribe an effective course of treatments.

The problem is that there are many people who don’t get their eyes tested every year, either because they choose not to or because they don’t have access to regular eye exams.

Google is trying to help more people get their eyes checked annually by using machine learning to screen people for diabetic retinopathy.

  • A research team at Google conducted a study in 2016 and found that, by utilizing machine learning to examine retinal photographs, eye specialists might be able to screen a lot more patients at once, including those who live in parts of the world where eye care is not always readily available.
  • It’s not a perfect system just yet, but the researchers at Google are eager to continue finding new applications for machine learning and artificial intelligence in healthcare settings.

In the future, machine learning could make it far easier for physicians to diagnose people with diabetic retinopathy and a host of other diseases as well.

Overdue for your annual eye exam? Call us today at 716-632-3545 to schedule an appointment or contact us here.