Scientists in the UK have developed an artificial intelligence (AI) system that can analyse eye scans taken during visits to an optician and use them to predict whether the person has a risk of developing cardiovascular disease.
In less than 60 seconds, it should be possible to predict whether a person is likely to have a heart attack within the next year, the researchers told The Guardian.
The tool has the potential to replace invasive blood tests and blood pressure checks that are currently used to screen for heart conditions, and also to test people outside clinics.
Vessel size as indicator
The AI looks for any tiny changes in the blood vessels in the retina. It assesses for heart attack risk from changes in the width, area, and curviness of vessels – all known indicators of vascular diseases, including problems in the heart.
The researchers automated deep learning algorithms to read the retinal scans
“AI-enabled vasculometry risk prediction is fully automated, low cost, non-invasive and has the potential for reaching a higher proportion of the population in the community,” the researchers said in a statement. They added that the technique can be made available more broadly without requiring additional testing.
Meets the standard score
Dubbed QUantitative Analysis of Retinal vessels Topology and siZe (QUARTZ), the system was trained to detect “stroke, heart attack, and death from circulatory disease” using retinal scans of 88,052 individuals aged 40-69 from the UK Biobank.
The researchers then validated QUARTZ by running it on retinal scans of 7,411 individuals aged 48 to 92 from another study – European Prospective Investigation into Cancer (EPIC) Norfolk. They said a risk score based on age, sex, smoking, medical history and retinal vasculature was performed as well as the Framingham Risk Score, a standard in assessing cardiovascular risk.
“In the general population it could be used as a non-contact form of systemic vascular health check, to triage those at medium-high risk of circulatory mortality for further clinical risk assessment and appropriate intervention,” the researchers added.
An `attractive option’
Dr Ify Mordi and Professor Emanuele Trucco from the University of Dundee, Scotland, who were not part of the study, said that using AI-based eye testing to predict heart disease risk was an “attractive and intuitive” option.
However, they added that more research needed to be done to determine whether the technique could improve clinical outcomes, and if so, workflows would need to be built to implement it practically.
