Scientists have developed a system in which artificial intelligence (AI) can deduce whether someone has an increased risk of heart failure from an eye scan.
Even the routine scans that people often have when visiting the optician or an eye clinic are sufficient to identify the risk of, for example, a heart attack. It’s the changes in the tiny blood vessels in the retina that doctors say are indicators of vascular disease, including problems with the heart.
The international research, in which KU Leuven also participated, was led by the University of Leeds in the United Kingdom. An AI system was developed to read retinal scans using deep learning techniques automatically. People who would have a heart attack with great certainty within a year could be identified in this way.
Deep learning involves a complex set of algorithms that allow computers to identify patterns in data and learn to make predictions. For example, with the newly developed AI system, the pumping efficiency of the left ventricle, one of the heart’s four chambers, can be estimated from the retinal scans. An enlarged ventricle may indicate an increased risk of heart disease.
This information, combined with basic patient data such as age and gender, allows the AI system to predict the risk of a heart attack over the next 12 months.
In Nature Machine Intelligence, the researchers write that the system has an accuracy of between seventy and eighty percent for detecting heart diseases.
Cardiovascular diseases, including heart attack, are among the most common causes of death worldwide. ‘This technique opens up the possibility of a revolutionary new way of screening’, says Scientific Director Alex Frangi of the University of Leeds. “The system may also work to detect heart disease at an early stage.”