Deep learning predicts, from retinal images, cardiovascular risk factors—such as smoking status, blood pressure and age—not previously thought to be present or quantifiable in these images.
