Artificial Intelligence Google has learned to predict the risk of heart disease in the eyes of the patient

Artificial Intelligence Google has learned to predict the risk of heart disease in the eyes of the patient

Researchers from Google and its subsidiary Verilu, specializing in medical technology, have created an algorithm capable of predicting the risk of a person developing cardiovascular diseases based on analysis of retinal pictures. An article on the development was published in the journal Nature Biomedical Engineering .

According to The Verge , the method proposed by researchers has approximately the same prediction accuracy as existing tests, for which it is necessary to take a blood test from a patient.

To train the algorithm, specialists used a database containing retinal images and medical indicators of almost 300,000 patients. On the basis of these data, the program has learned to determine the age of a person, its blood pressure from images of the retina, and whether it is a smoker. Together, this information allows you to judge the risks of developing heart disease.

During the test, the algorithm comparing the retinal pictures of two patients, one of whom experienced heart problems within five years after the shot was taken and the other not, correctly determined each patient in 70% of cases. At the same time, the currently used method of SCORE in a similar comparison of blood tests allows predicting the development of cardiovascular diseases with an accuracy of 72%.

The Verge notes that in the long term, the use of the Google algorithm can simplify the work of doctors, but first this method of identifying heart diseases should be more carefully checked.