A research team found that combining information about the pattern of blood vessels in the retina with conventional clinical variables resulted in a more accurate identification of heart attack risk than previous models based only on demographic information. In summary: An eye test can make heart attack risk predictions more accurate.
In an abstract To be presented today at the annual conference of the European Society of Human Genetics in Vienna, the researchers explain how they used data from the British biobank (medical and lifestyle information from 500.000 people) to come up with a metric known as the fractal dimension.
They then combined these data into a model with factors such as age, sex, systolic blood pressure, body mass index, and smoking status, and combined these factors with images of patients' retina obtained from an eye exam.
A more accurate model
Ana Villaplana Velasco, author of the study and PhD student at the Usher and Roslin Institutes of the University of Edinburgh, is enthusiastic. “Surprisingly, we found that our model classified heart attack risk better than established models. Our analysis found that there is a shared genetic basis between the fractal dimension and myocardial infarction.”
The average age of someone who suffers a heart attack is 60: and this algorithm has shown an ability to predict it even five years earlier. The previous record of an AI developed by the University of Leeds was one year old. Researchers believe that in the near future, a simple eye exam can provide adequate information to identify those at risk.
“An individualized exam would allow doctors to suggest specific behaviors to anyone over 50,” Villaplana-Velasco says.
An eye test to observe... the "eyes of the heart"
According to the researchers, the retinal variant profile may be unique to each of us, helping to identify all types of medical conditions. Therefore, their discovery could help calculate the risk of contracting other diseases as well. The next step, however, is to evaluate whether an eye exam can help produce two specific models (one male and one female) to assess the risk of heart attack even more accurately.
More research is needed to show that this forecast improvement is robust. It takes work to understand the feasibility of this approach and understand how to build a valid routine.
Although this abstract still offers limited data, the retina (it is now established) will give us unique opportunities to evaluate vascular health. Using machine learning to store large amounts of data and/or correlations could allow us in the future to do an entire medical check-up with just a quick eye test.
