UK Biobank participants have played a critical role in a new piece of research that might make it easier to predict heart attacks and strokes in vulnerable people.
The work, led by Dr Lily Peng, Google Research brain team, used pictures of participants’ eyes, taken when they joined the study, to show that images can accurately predict other indicators of cardiovascular health.
The research is particularly interesting because it offers the possibility of finding more ways to diagnose health issues from eye images, and might one day reduce the need for more invasive tests.
Dr Peng said: “We used the UK Biobank data to train deep learning models to make predictions from retinal fundus images about cardiovascular risk factors and a 5-year risk of a major cardiovascular event, such as a heart attack or stroke.
“We discovered that by leveraging UK Biobank’s large dataset, we were able to train a model that could very accurately predict genetic factors associated with cardiovascular disease such as age and sex, lifestyle factors such as smoking and directly predicting a major cardiovascular event.
“UK Biobank was critical to this research taking place, serving as one of the basis of these discoveries.”
You can read more about Dr Peng’s work here: Assessing Cardiovascular Risk Factors with Computer Vision
A copy of the research paper can be found via the link below: