THE ESSENTIAL
- Study finds retinal photograph can predict stroke risk using artificial intelligence-based model.
- This tool analyzes 29 retinal indicators, such as vascular density and complexity, and offers accuracy comparable to traditional methods without invasive testing.
- By studying more than 45,000 participants, researchers showed that certain specific retinal variations increased the risk of stroke by 10 to 19%.
What if a simple photograph of the retina could assess your risk of stroke? In a study recently published in the journal Heartresearchers presented an innovative model that predicts stroke with accuracy comparable to conventional methods, but without the need for invasive testing.
A unique vascular footprint
The retina, the thin layer of tissue at the back of the eye, has a complex vascular network sharing many anatomical and physiological similarities with the brain. This connection makes it a valuable tool for detecting damage caused by systemic health problems like diabetes. However, its potential for assessing stroke risk has until now been underexploited.
The study is based on an analysis system based on artificial intelligence, the Retina-based Microvascular Health Assessment System (RMHAS). This tool uses fundus photography, a non-invasive technique, to measure 29 indicators of vascular health, such as the density, complexity and caliber of retinal veins and arteries.
Towards more accessible medicine?
The researchers analyzed the retinal images of more than 45,000 participants with an average age of 55, we can read in a press release. During the 12.5 years of follow-up, 749 of them suffered a first stroke. The usual risk factors were present: advanced age, diabetes, smoking and even hypertension. But more importantly, the study found that variations in certain retinal indicators were strongly associated with an increased risk of stroke. For example, each change in vascular density increased this risk by 10 to 19%, while changes in vascular caliber were associated with an increased risk of 10 to 14%.
According to the authors, combining retinal images with simple data such as age and gender offers an effective method for predicting stroke. This tool, easy to use in a doctor’s office, could expand access to early diagnosis and revolutionize prevention, as strokes affect nearly 100 million people each year.
Promising and easy to use in primary care, this method however requires additional studies to confirm its effectiveness in more diverse populations, temper the authors of the study.
