A new AI system can scan routine eye exam images to estimate a person’s risk of heart disease and stroke, closely matching standard risk calculators. Researchers say the approach could turn eye clinics into powerful checkpoints for cardiovascular prevention.
A quick snapshot of the back of your eye could soon do more than check your vision. It might also flag your risk of heart attack or stroke.
Researchers have developed an artificial intelligence system that analyzes retinal images taken during routine eye exams and estimates a person’s cardiovascular risk. In a new multicenter study, the tool’s results closely tracked a widely used heart risk calculator based on blood pressure and cholesterol.
The findings, presented at the American College of Cardiology’s Annual Scientific Session (ACC.26), suggest that eye clinics could become an unexpected front line for heart disease prevention.
Heart disease remains the leading cause of death worldwide. Doctors typically use risk calculators to estimate a person’s chances of developing atherosclerosis, the plaque buildup in arteries that can lead to heart attacks, strokes and early death. Those estimates help guide advice on lifestyle changes and medications such as statins.
But many people do not see a primary care provider regularly and may have no idea they are at elevated risk.
“The awareness that someone might be at risk is really one of the key missing pieces,” lead author Michael V. McConnell, a clinical professor of medicine at Stanford University, said in a news release.
McConnell is also chief health officer at Toku, the company that developed the AI system, called CLAiR. The software has received Breakthrough Device designation from the U.S. Food and Drug Administration, and this first prospective U.S. evaluation is intended to support a formal FDA submission.
The idea behind CLAiR is simple but powerful: use the eye as a window into the body’s blood vessels.
“Even just a standard retinal photo provides high resolution imaging of your blood vessels—it’s a literal window into vascular tissue,” McConnell added.
The back of the eye, or retina, is lined with tiny arteries and veins that reflect the health of the broader vascular system. Previous research has shown that eye images can reveal signs of conditions such as diabetes and high blood pressure, but those assessments have typically depended on human experts.
CLAiR aims to automate that process. The AI was trained to recognize patterns in retinal blood vessels that are associated with future cardiovascular problems. Once a retinal photo is taken, the system analyzes the image and estimates whether a person’s 10-year risk of heart disease or stroke is high enough that they would likely benefit from preventive medications.
In the new study, researchers enrolled 874 adults ages 40 to 75 who were not taking cholesterol-lowering drugs and did not have known atherosclerosis. Participants were recruited from 10 eye care and primary care sites across the United States. About half were women, and the group included Black or African American and Hispanic participants.
Each person had retinal images taken with standard cameras commonly found in eye clinics. The CLAiR system then analyzed the photos and flagged participants whose estimated 10-year risk of heart disease or stroke was 7.5% or higher, a common threshold for recommending statin therapy.
At the same visit, clinicians also collected traditional risk factors — age, sex, smoking status, blood pressure and cholesterol — and calculated each participant’s 10-year risk using the standard atherosclerotic cardiovascular disease (ASCVD) risk estimator.
Overall, 26% of participants had a 10-year ASCVD risk score of 7.5% or greater. CLAiR’s performance closely matched that benchmark. The AI correctly identified people in the elevated-risk group 91.1% of the time (sensitivity) and correctly recognized those not in the elevated-risk group 86.2% of the time (specificity).
Those numbers suggest the system could be a reliable first-pass screen in eye care settings, helping to catch people who might otherwise slip through the cracks.
“The image of the back of your eye has a wealth of health information. We can analyze these images with AI to help people become aware of their risk and have the opportunity to get guideline-based evaluation and preventive therapy,” added McConnell.
The study also showed that the approach is practical in real-world clinics. Ninety-four percent of the retinal images collected were usable by the AI system, even though they came from different cameras in different locations. Retinal imaging took about five minutes, and CLAiR returned results in about 30 seconds, meaning the process added little time to a typical visit.
The researchers emphasized that the tool is not meant to replace a full cardiovascular workup. Instead, it could serve as an early warning system that prompts people to seek further evaluation from a primary care clinician or cardiologist.
“This approach would not replace the standard cardiovascular risk evaluation, but it’s a potential way to bring greater awareness, especially for people who should be on preventive care, but who have not yet had a thorough evaluation,” McConnell said.
Turning that potential into better health outcomes will require more than just accurate AI. Clinics will need systems to ensure that people flagged as high risk actually get follow-up care.
“For patients to benefit, we need to implement clear pathways to connect your elevated risk from your eye exam to help you see your clinician and ultimately get guideline-based preventive therapy,” McConnell added.
There are also practical barriers. While retinal imaging equipment is available in most U.S. eye clinics, the procedure is not covered by all vision insurance plans as part of a standard visit. Patients may have to pay an extra fee, which could limit access.
The CLAiR system is not designed for everyone. It is not intended for use in people who are pregnant or those with advanced eye disease, which can alter the appearance of retinal blood vessels.
The study, formally titled “Prospective Multi-center Clinical Trial of Artificial-Intelligence Analysis of Retinal Images for Identifying Elevated Atherosclerotic Cardiovascular Risk,” was funded by Toku. McConnell presented the results March 30 at ACC.26 in New Orleans, where cardiologists and cardiovascular specialists from around the world gathered to share new advances in heart disease treatment and prevention.
If regulators ultimately clear CLAiR for clinical use, eye exams could become a powerful new touchpoint for catching silent cardiovascular risk — offering patients a chance to protect their hearts at the same time they protect their sight.
Source: American College of Cardiology