A new system that uses artificial intelligence (AI) to assess cardiovascular risk based on images of the eyes taken during eye exams demonstrated strong correlations with standard cardiovascular risk assessments, according to research presented at the American College of Cardiology’s Annual Scientific Sessions (ACC.26). Researchers said using AI to screen for heart disease risk during routine eye exams could make more people aware of their risk and encourage referrals for preventive treatment.
Heart disease is the leading cause of death worldwide. Primary care providers and cardiologists use risk calculators to estimate a patient’s risk for atherosclerosis (plaque buildup in the arteries that can lead to heart attack, stroke, and premature death) and guide lifestyle modification and medication recommendations as needed. However, not everyone sees their primary care provider regularly and may not be aware of their risks or the steps they can take to reduce them.
Recognizing that someone may be in danger is actually one of the key elements that is missing. Images of the back of the eye contain a wealth of information about health. Analyzing these images with AI can help people become aware of their risks and give them the opportunity to receive guideline-based assessments and preventive treatments. ”
Michael V. McConnell, MD, clinical professor of medicine at Stanford University, Stanford, California, and lead author of the study
McConnell is the chief health officer of Toku, the company that developed the AI system used in the study. The system, called CLAiR, received breakthrough device designation from the U.S. Food and Drug Administration (FDA). The results of this first prospective evaluation of CLAiR in the United States will support submission to the FDA.
The study found that one in four participants had an increased risk of heart disease, based on standard cardiovascular risk assessments such as blood pressure and cholesterol screening. An AI-based method that used retinal images taken during the visit to analyze blood vessels at the back of the eye closely matched this determination, identifying at-risk participants with a sensitivity of 91.1% and specificity of 86.2%.
“Standard retinal photography alone provides high-resolution images of blood vessels; it’s a literal window into vascular tissue,” McConnell says.
Previous research has shown that images of the eye can be used to assess conditions such as diabetes, but most methods have relied on interpretation by human experts. With CLAiR, the developers sought to demonstrate how this approach could be scaled up for clinical implementation by using AI to automate image analysis. The AI system was trained to recognize patterns in the appearance of blood vessels that are associated with the development of heart disease.
Researchers enrolled 874 participants between the ages of 40 and 75 who were not taking lipid-lowering drugs and had no known atherosclerosis. Participants were recruited at 10 eye care and primary care facilities across the United States. Half were women, 19% were black or African American, and 26% were Hispanic.
Each participant underwent retinal imaging with a standard camera used in most eye clinics. The images were then analyzed using the CLAiR system to identify participants who had a 7.5% or higher chance of experiencing heart disease or stroke in the next 10 years. This is a commonly used threshold to identify patients who are likely to benefit from taking statins.
Standard ASCVD risk estimation tools were also used to identify participants in the 7.5% 10-year risk category for comparison with CLAiR results. To this end, data on participants’ age, gender, smoking status, blood pressure, and cholesterol were collected during the same clinic visit.
Using standard risk estimators, a total of 26% of participants were found to have a 10-year ASCVD risk score of 7.5% or higher. The CLAiR system was consistent with these results, correctly identifying positive cases 91.1% of the time (sensitivity) and negative cases 86.2% of the time (specificity).
Based on their results, the researchers said that while AI systems show promise as a viable non-invasive screening method in most eye care settings, further efforts are needed to facilitate the referral of at-risk patients for cardiovascular evaluation and treatment in primary care after retinal imaging screening.
“While this approach is not a replacement for standard cardiovascular risk assessments, it could be a way to bring greater awareness, especially for people who should receive preventive care but have not yet been adequately assessed,” McConnell said. “For patients to benefit, we need to put in place clear pathways to help them identify elevated risks with eye exams, seek medical advice and ultimately receive guideline-based preventive treatment.”
Overall, 94% of the images captured in the study were usable by the AI system, providing evidence that the approach works well across cameras used in a variety of clinics. Retinal imaging takes about 5 minutes, but the CLAiR algorithm returns results in about 30 seconds. This suggests that implementation of this approach does not add much time to the clinical workflow. McConnell said the CLAiR system is not designed for pregnant women or people with advanced eye diseases that can affect the condition of the blood vessels in the eye.
Although retinal imaging is available at most eye clinics in the United States, it is not covered by all vision insurance plans as part of a standard visit, and patients may be charged extra for imaging.
This research was funded by Toku, the developer of CLAiR.
Dr. McConnell will present his study, “Prospective Multicenter Clinical Trial of Artificial Intelligence Analysis of Retinal Images to Identify Elevated Atherosclerotic Cardiovascular Risk,” on Monday, March 30th at 10:45 AM CT/3:45 PM UTC at La Nouvelle B.
sauce:
American College of Cardiology
