Google (NSDQ:GOOG) parent Alphabet‘s (NSDQ:GOOGL) Verily is developing a new artificial-intelligence powered test that searches for indicators of heart disease risk present in retina images, according to a new study.
The study, which has not yet been peer reviewed but has been published by Cornell University, details a new machine-learning powered system designed by Verily to “discover new knowledge from retinal fundus images.”
The system operates on data from 284,335 patients, validated on 2 independent datasets of 12,026 and 999 patients, to predict cardiovascular risks “not previously thought to be present or quantifiable in retinal images,” according to the study publication.
Researchers claim they were able to use the system to determine a patient’s age accurate to within 3.3 years, gender, smoking status, HbA1c levels, systolic blood pressure and major adverse cardiac events from retinal images.
The AI-powered system uses “distinct aspects of the anatomy to generate each prediction,” according to the study, which researchers claim could open up avenues of further research into the technique.
Study investigators touted that the system was comparable to the European SCORE risk calculator, but offered the caveat that study came with several limitations, including a “relatively small” dataset size for deep learning.
“In particular, although the AUC for cardiovascular events was comparable to SCORE, the confidence intervals for both methods were wide. A significantly larger dataset or a population with more cardiovascular events may enable more accurate deep learning models to be trained and evaluated with high confidence,” study authors wrote.
Researchers concluded that the study “provides evidence that deep learning may uncover additional novel signals in retinal images that will allow for better cardiovascular risk stratification,” and were supportive of further research of such techniques.