Verily researchers develop AI algorithm to detect diabetic retinopathy

Researchers from Verily, Alphabet's life sciences arm, developed a deep learning algorithm that can identify diabetic retinopathy with equal or better accuracy than that of trained human experts, according to a study published this month in JAMA Ophthalmology.

The study was conducted at two eye care centers in India, Aravind Eye Hospital and Sankara Nethralaya, where more than 3,000 diabetic patients were screened. The artificial intelligence, which has been in development by Verily since 2016, identified diabetic retinopathy with nearly 90 percent accuracy, a rate equal to or, in some cases, exceeding that of trained graders and retina specialists.

"The automated system holds promise as a point-of-care initial screening solution that does a first pass to rule out patients at lower risk of vision-threatening disease and flagging images that are categorized as abnormal for timely follow-up by a clinician," the study's authors wrote. "This will decrease the proportion of patients that might be lost to follow-up because of a failure to return the test results to the patient asynchronously or the need for repeated visits. This would be especially advantageous in low-resource settings."

More articles about AI:
Amazon Echo and other voice assistants can be trained to hear cardiac arrest
IBM AI predicts breast cancer up to a year in advance using health records, mammograms
Discharge algorithm could save hospitals $860 per care episode

© Copyright ASC COMMUNICATIONS 2019. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Top 40 Articles from the Past 6 Months