ADA25: Mediwhale’s AI pairs retinal scans, metabolites to refine diabetes risk prediction

Mediwhale is making the case for using retinal scans to predict heart and kidney disease risk. Data presented at the American Diabetes Association (ADA) annual meeting suggest that its AI tool, known as Dr. Noon, performs on par with standard clinical models and can detect risk differences not picked up by conventional tests. The platform analyses retinal images to provide a non-invasive alternative for identifying high-risk individuals, including those with diabetes or prediabetes.

At the meeting, researchers presented two studies using data from the UK Biobank. One focused on 10-year cardiovascular risk, the other on five-year risk of chronic kidney disease (CKD) in people with diabetes. Both compared Dr. Noon to established clinical models and tested whether a multi-omics approach combining retinal imaging with blood metabolite data could improve predictive performance.