Researchers from Google, Verily Life Sciences, and Stanford School of Medicine have developed an algorithm to predict cardiovascular risk factors from retinal fundus photographs. Their model was trained from 48,101 patients from the UK Biobank and 236,234 patients from EyePACS using deep learning, and validated on 12,026 patients from the UK Biobank and 999 patients from EyePACS. It is able to predict cardiovascular risk factors not previously thought to be present or quantifiable in retinal images, such as age, gender, smoking status, systolic blood pressure and major adverse cardiac events.

The work was published in Nature Biomedical Engineering. For more details, please visit https://www.nature.com/articles/s41551-018-0195-0.