Running eye scans through AI algorithm ‘accurately predicts risk of an early death’

Scanning the inside of someone’s eye could predict their risk of an early death, a study suggests.

Scientists have developed an AI programme that can spot early warning signs of a heart problem.

It works by measuring the thickness of tiny arteries and veins at the back of the eye, which are thought to hold key information about heart health.

Problems with blood circulation can cause blood vessels in the retina to thicken eventually leading to vision loss without treatment.

In a study of 88,000 Britons, it was able to predict 54 per cent of cases of heart attacks and strokes and 58 per cent of heart disease deaths.

Doctors said the test offered an alternative that was ‘non-invasive’ and could be used to check heart health during regular eye check ups.

Pictured above is a retina (in the black box) and a cut out showing how the AI measures the thickness of an artery running across its surface. A new method has been developed using these to suggest how likely someone is to die from heart disease

Heart disease is the leading cause of death in America claiming about one life every 34 seconds, or 690,000 people annually.

There are no guidelines on how often someone should be screened for heart disease.

But the Centers for Disease Control and Prevention recommends all over-20s are screened for high cholesterol — a predictor of heart disease — at least every five years. 

They also recommend regular tests for blood pressure, and checks for blood glucose levels every three years.

The tests are typically done by taking into account age, sex, smoking status and blood pressure to determine risk from heart disease. Doctors also look at cholesterol levels so they require a blood test.

The new method used retinal vasculature imaging, which photographs the back of the eye to show blood vessels along with the optic nerve.

It is used by doctors during eye tests to check for health issues, such as degeneration of the retina — and glaucoma where the optic nerve is damaged.

They developed a computer program that can read the scans to take thousands of measurements of the width of blood vessels across the retina.

It then analyzes these to see if they have thickened.

The results are then used alongside someone’s age, smoking status and medical history, to estimate their risk of death from heart disease.

Results showed that in the British sample the algorithm predicted 191 deaths from heart disease, but 327 were recorded (58 per cent accurate).

It also estimated there would be 454 cases of strokes or heart attacks in the group, and 839 were recorded (54 per cent).

Follow-up for the British sample was about seven years after the initial eye exam on average.

The algorithm was also tested on a separate group of 7,400 patients in Europe aged between 48 and 92 years.

It found 50 per cent of the 201 deaths from heart disease recorded, and 48 per cent of the strokes and heart attacks recorded over a period of about nine years.

Dr Alicja Rudnicka, an epidemiologist who led the paper, and others said it gave a ‘better prediction’ for risk from heart disease than current methods.

They added: ‘In the general population it can be used as a non-contact form of systemic vascular health check, to [move] those at medium-high risk… for further clinical risk assessment.’

In a linked editorial, Dr Ify Mordi and Dr Emanuele Trucco, clinicians at Dundee University in Scotland, said the new method was ‘attractice’ to use.

‘Using retinal screening in this way would presumably require a significant increase in the number of ophthalmologists or otherwise trained assessors,’ they suggest.

‘What is now needed is for ophthalmologists [and others] to work together to design studies to determine whether using this information improves clinical outcomes.’

The method — named QUantitative Analysis of Retinal vessels Topology and SiZe (QUARTZ) — was revealed in a study in the British Journal of Ophthalmology.