Retinal Scans to Predict Risk of Cardiac Infarction

Article: AI can identify heart disease from an eye scan
Source: University of Leeds (U.K.)
Published: January 25, 2022
Article: Eye Scans Open a Window into Heart Health
Source: Genetic Engineering & Biotechnology News
Published: January 26, 2022
Article: A Hidden Pattern in Your Retina May Reveal if You're at Risk of a Future Heart Attack
Source: ScienceAlert
Published: January 27, 2022

Graphical abstract

The health of the blood vessels in the retina is an indication of systemic health, such as in diabetes and vascular disease. Given the connection between this ocular window to cardiovascular health, an international team of researchers explored identifying patients at risk of myocardial infarction based on their retinal scans, which differs from earlier similar projects that used retinal images and deep learning to predict various diseases, but not cardiac events such as heart attack. The scientists used data from 70,000 individuals from the U.K. Biobank to train a computer deep learning algorithm, and further tested the model on a database of 3,000 individuals in the U.S. AREDS cohort. The A.I. algorithm then applied a multi-channel variational autoencoder and a deep regressor model to a data set of 5,663 people to analyze associations between retinal scans and cardiac scans (mass and end-diastolic volumes) of a patient's left ventricle, one of the four chambers of the heart, to estimate the heart's pumping efficiency; an enlarged ventricle indicates higher risk of heart disease. This deep learning system could predict if patients were at risk of a heart attack over the next year with a 70-80% accuracy, the researchers report. One of the researchers explains, "Our method can use imaging of both the fundus and the heart during the training phase yet only require fundus images during risk inference." Given the noninvasive, accessible, and affordable nature of retinal imaging, this new method could be used as an early screening of heart health in place of echocardiography or magnetic resonance imaging of the heart. The research team looks to further study confounding factors and to test their model in clinical trials of the general population in the future.

My rating of this study: ⭐⭐⭐

Diaz-Pinto A, Ravikumar N, Attar R, et al. "Predicting myocardial infarction through retinal scans and minimal personal information." Nature Machine Intelligence.  25 January 2022. https://doi.org/10.1038/s42256-021-00427-7