Researchers at Leeds University have developed an AI algorithm that can detect early signs of heart failure from routine retinal scans, potentially saving millions of lives through early intervention.
A groundbreaking study published in Nature Medicine demonstrates how deep learning models applied to retinal fundus images can predict future cardiovascular events with remarkable 80% accuracy, potentially revolutionizing preventative cardiology worldwide.
The research team at the University of Leeds, led by Professor Alex Sheridan, trained their AI model on over 500,000 retinal images paired with 10-year cardiovascular outcome data from the UK Biobank. The algorithm identifies subtle vascular changes in the retina that precede clinical symptoms by years or even decades.
"The eye is a unique window into the body's vascular health," explained Professor Sheridan. "The retinal blood vessels share many characteristics with coronary arteries, but they're directly visible without invasive procedures. Our AI can detect changes that are invisible to the human eye."
The model analyzes over 200 different features in each retinal image, including vessel diameter, tortuosity, branching patterns, and subtle color variations in the retinal tissue. It then generates a cardiovascular risk score that outperforms traditional risk calculators like the Framingham Risk Score.
In validation studies across three continents, the AI correctly identified 84% of patients who would experience a major cardiac event within 10 years, while maintaining a low false-positive rate of just 12%. This performance was consistent across different ethnic groups and age ranges.
The implications for global health are substantial. Retinal imaging is already widely available in optometry practices worldwide, and the AI can run on standard computing hardware. This means cardiovascular screening could be integrated into routine eye exams at minimal additional cost.
The NHS has announced a pilot program to implement the technology across 50 primary care centers in England, with plans for broader rollout pending the results. Several private healthcare systems in the United States and Singapore have already begun integration.
"We envision a future where a simple eye exam during a routine optician visit could identify people at high cardiovascular risk, enabling early intervention with lifestyle changes and preventative medications," said Professor Sheridan. "This could prevent millions of heart attacks and strokes globally."