Most people rely on how advanced artificial intelligence is. There is a published study from the American Academy of Ophthalmology which shows that when artificial intelligence (AI) and physicians work together, it is more effective in diagnosing diabetic retinopathy than doing it alone.The study focused initially on the comparison of performance between the AI system and human specialists. The study aims to investigate how AI can be used in the real world. They wanted to see if the doctor performed better alone versus the algorithm alone or would the combination of a doctor and an algorithm be better when it comes to diagnosis.
The study included 10 ophthalmologists from different training and experience. They will grade from normal to severe the 1,796 eye pictures of diabetic patients. The images will be graded by the ophthalmologists without the algorithm, with the algorithm, and with the combination of algorithm and an explanation of why. The two types of assistance improved the accuracy and confidence of the physician in the diagnosis. The research highlighted the second opinion effect of AI because they perform like another doctor.
The Role of Artificial Intelligence in Diagnosing Diabetic Retinopathy
Based on the statistics, more than 29 million Americans have diabetes and are at risk for diabetic retinopathy. Diabetic retinopathy is a potentially blinding eye disease that needs to be detected and treated early. The earlier, the better because it can prevent or minimize vision loss. It is recommended for people with diabetes to have yearly screenings but unfortunately, not all people have access to an ophthalmologist.
Be reminded that the accuracy of screening exams also varies among medical professionals significantly. A study showed a 49 percent error rate in diagnosing advanced diabetic retinopathy among medical residents, internists, and diabetologists. It has been a challenge to provide an accurate screening for millions of people who are diabetic. An AI system was designed to analyze and interpret pictures of the eye like a human grader or doctor would. It was trained to detect the difference between images of normal or healthy eyes and abnormal or unhealthy eyes.
Can We Rely on AI for an Accurate Eye Screening?
There have been studies where AI software uses deep learning techniques in identifying early warning signs of diabetic retinopathy. The first AI-based device that is approved by the U.S. Food and Drug Administration is the IDx-DR which is used in primary care settings to identify patients who need a referral to an eye specialist.
The assistance of AI for eye screening is just getting started. It will take a long way to go before we can solely depend on AI. Currently, the idea of AI systems working with human intelligence can improve patient care.