Machine learning can be used to predict a person's risk of heart disease in less than a minute by analyzing veins and arteries in their eye.
If the findings of the new research are found to be valid in future trials, it will pave the way for the development of inexpensive cardiovascular screenings. The screenings would let people know if they have a stroke or heart attack.
Alicja Rudnicka told The Guardian that the tool could let someone know in 60 seconds their level of risk. The study found that the predictions were correct.
It is possible to use the eye as a window.
Blood vessels in the eye are analyzed by the software. The total area and width of the arteries and veins are measured. The software is able to make predictions about a subject's risk from heart disease just by looking at a non-invasive snapshot of their eye.
The eye can be used as a window to the rest of the body according to a researcher not connected to the study. Doctors have known for a long time that you can see signs of high blood pressure and diabetes by looking in the eye. Manual assessment was the problem. The use of machine learning can make a difference.
Machine learning medicine uses artificial intelligence to diagnose diseases from eye scans. The first ever artificial intelligence device approved by the FDA was used to screen for eye disease, and research shows that it can detect a range of ailments in this way. Questions about the reliability and universality of their diagnoses are still being asked of the tools applying these findings.
The study was only done on the eye scans of white patients. The UK Biobank contains test data that is 94.6% white and reflects the UK's own demographic in age range of patients included in the bank. The biases need to be balanced in the future to make sure the tool is accurate.
The researchers compared the results from their software, named QUARTZ, with 10-year risk predictions made by the standard Framingham Risk Score test The two methods had similar performance.
The challenge is taking this type of work to a clinic. He wondered if the UK's National Health Service or a company spun out from the university could turn this type of research into a diagnostic tool. What level of performance will be required by regulators before they approve the software? We should put a fork in it and make it into a commercial product.
All Rights Reserved © 2024
