In new advances in science, researchers have found that automated; The Artificial Intelligence (AI) screening system can accurately detect diabetic retinopathy 95.5 percent of the time.
In addition, the system does not require inputs from a specialist ophthalmologist and provides a reading in just 60 seconds. Researchers presented these findings at AAO 2019, the 123rd annual meeting of the American Academy of Ophthalmology.
Diabetic retinopathy develops over time in people with diabetes, especially when there is no proper control over their blood sugar levels.
Excessive blood sugar can damage the small blood vessels in the back of the eye. Sometimes, small bulges protrude from the blood vessels, leaking fluid and blood into the retina.
This fluid causes inflammation or edema in the area of the retina, allowing it to be seen clearly. First, diabetic retinopathy causes no symptoms or only mild vision problems. In the end, it causes blindness. Ophthalmologists have effective treatments for diabetic retinopathy, but they work best when the condition is initially diagnosed. That is why ophthalmologists recommend examining patients every year. A system called IRT has shown promise in previous studies. It was used to test 893 patients with diabetes at 15 different medical centers. Results are reviewed for clinical accuracy by certified graders. The sensitivity of the IRT system was 95.5 percent, and the specificity was 86 percent, with patients’ students not disentangled.
Only a small fraction of the eyes need dilation to achieve an image that is good enough to grade. When these additional patients were included in the analysis, the sensitivity was the same, the specificity improved to 86.5%, and the grade ability improved to 97.4%.
More than 90% of the eyes identified as positive by the IR system have diabetic retinopathy or another eye disease according to the reference standard.