Connect with us

Hi, what are you looking for?

Tech & Science

Eye imaging predicts cardiovascular risk

Trials suggest that Google’s new algorithm can predict cardiovascular risk from scanning and interpreting eye images. This follows on from other eye-related research a Google where deep learning techniques have successfully predicted eye-diseases as well as other conditions based on eye-scans, such as diabetic eye disease.

The diabetes related condition was presented in the article “Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs“, which was published in the medical journal JAMA. The combination of scanning and we the deep learning algorithm was based on an assessment of retinal photographs.

With the extension of the technology to hear conditions, Google trained its algorithm on data relating to 284,335 patients. Following this exercise, Google states that its machine learning intelligence was able to predict cardiovascular risk factors from the scans of retinal images. This performance was able to match the test accuracy of alternate cardiovascular risk calculators that need a blood test, such as tests for cholesterol which correlate to heart attack risks.

Other information of medical interest could also be garnered. The algorithm was also, for example, able to predict the systolic blood pressure within 11 mmHg on average for patients overall.

Commenting on the overall performance, Dr. Lily Peng, who is the Product Manager at Google Brain Team, told Smart2Zero: “At the broadest level, we are excited about this work because it may represent a new method of scientific discovery.”

The researcher adds: “Traditionally, medical discoveries are often made through a sophisticated form of guess and test — making hypotheses from observations and then designing and running experiments to test the hypotheses.”

Overall the deep learning algorithm was able to make connections between changes in the human anatomy of the eye and equate these to the likelihood of disease. This is similar to the way that a medical student learns how to associate signs and symptoms with medical diagnosis and disease prediction.

The algorithm is not yet ready for medics to use. Google plans to assess the algorithm on bigger and more comprehensive datasets, which include cases of patients who suffered cardiovascular events.

The research to date has been published in the journal Nature Biomedical Engineering, with the research titled “Prediction of Cardiovascular Risk Factors from Retinal Fundus Photographs via Deep Learning.”

Written By

Dr. Tim Sandle is Digital Journal's Editor-at-Large for science news. Tim specializes in science, technology, environmental, and health journalism. He is additionally a practising microbiologist; and an author. He is also interested in history, politics and current affairs.

You may also like:


Some hardline Hindu groups are claiming the famed Taj Mahal was built on the site of a shrine to Shiva - Copyright AFP Yasuyoshi...


Unexploded ordnance is seen on a road in Severodonetsk, eastern Ukraine - Copyright AFP Ishara S. KODIKARAHere are the latest developments in the war...


UN data estimates that more than 620,000 Ukrainian refugees have crossed into Hungary since the invasion on February 24.


Russian troops abandoned positions near Kharkiv in haste when Ukrainian forces pushed them back towards the border - Copyright AFP PRAKASH MATHEMAOn a small...