A heart attack can occur when least expected, but researchers at the University of Leeds in the UK believe that an AI can predict when the attack will occur, reports the IEEE Spectrum. And that by just looking at pictures of your retina, which can be taken during a regular eye examination. The assumption is that small changes in the blood vessels in the retina can be a sign that a person is at high risk of having a heart attack soon.
The researchers have trained an AI to recognize the signs above by using machine learning. In this project, historical data from over 75,000 people were employed. The AI highlighted the cases where the risk of heart attack was high, and the result shows that AI was correct in 70-80 percent of the issues.
Compared to just looking at a person's general risk factors, the AI's results were, in fact, ten percent better. The AI got about the same results as a thorough heart examination, but just taking a retina picture is much cheaper than having a doctor do a comprehensive exam.
This procedure could become routine when you receive an eye examination in the future. The AI could quickly and cheaply determine which patients need to see a doctor and get a more in-depth heart examination.
So far, this is a preliminary study, and the researchers are now working on refining the method.