Health & Wellness

Google Knows How to Scan Our Eyes for Heart Disease

Research is giving a whole new meaning to getting your eyes checked

By Katrina Caruso


Did you know that Google owns a health science company named Verily Life Sciences? Google and Verily scientists have been looking at ways to scan for heart disease risk using the same sort of computer science that makes Google searches possible.

Using data collected from retinal scans, the new Google software was able to predict a patient’s age (within a margin of 3.26 years), blood pressure (within 11 units of the upper number), whether he or she was a smoker (with an accuracy of 71%), and more. The process may eventually be as precise as a blood test and provide results more quickly.

The study results were published in Nature’s journal Biomedical Engineering in January 2018.

The scientists used 284,335 patients’ eye scans (and other datasets) to create a deep-learning algorithm. About 48,000 patients were from the UK Biobank database, with the remainder from EyePACS medical database. With this data, they undertook two tests, involving 12,026 and 999 patients. The Google software compared two images from two different people, one who had had a cardiovascular event such as a heart attack within the last 5 years and one with no history of heart trouble: 70% of the time, the software was able to tell which patient had heart problems. That number may not sound very impressive, but it’s actually only a bit less accurate than the SCORE method (using a blood test for analysis), which is one of the commonly used methods for predicting cardiovascular risk.

There’s much more information still needed, as Google and Verily haven’t been able to test the method in a clinical setting. But the results so far are promising.

You know what they say about the eye being a window to the soul. Soon it might provide the best way for your doctor to check out your heart.

Photo: iStock/ferrantraite.