Researchers at Mass General Brigham, in collaboration with the U.S. Department of Veterans Affairs (VA), have developed an artificial intelligence tool that can scan existing chest CT images to detect high coronary artery calcium (CAC) levels, an important predictor of future heart attacks and long-term mortality. Their findings, published in NEJM AI, show that the tool, called AI-CAC, accurately identifies individuals at elevated cardiovascular risk and may help clinicians intervene earlier to prevent major cardiac events.
Millions of chest CTs are performed annually, often for reasons unrelated to heart disease, such as lung cancer screening. This study demonstrates that valuable cardiovascular information is hiding in plain sight, and AI can help unlock it.
Chest CT scans can reveal calcium buildup in coronary arteries, but the gold-standard method for CAC scoring requires “gated” CT scans synchronised with the heartbeat. Most routine clinical CTs, however, are “nongated” and not typically used for heart disease screening. The research team recognised that CAC is still visible on these nongated scans and developed a deep learning model—AI-CAC—to analyze them and quantify risk.
The AI-CAC model was trained using CT scans collected from veterans receiving routine care at 98 VA medical centers. It was tested on 8,052 scans to simulate real-world screening conditions. The model was 89.4% accurate in detecting the presence of CAC and 87.3% accurate in identifying whether a patient’s score exceeded 100, a marker of moderate cardiovascular risk.
AI-CAC also predicted 10-year all-cause mortality: individuals with CAC scores over 400 had a 3.5-fold higher risk of death compared to those with a score of zero. Cardiologist review confirmed that 99.2% of patients flagged by the tool as having very high CAC would be candidates for lipid-lowering therapy.
There are millions of nongated chest CTs in the VA system alone, compared to only about 50,000 gated scans. AI-CAC represents a powerful opportunity to repurpose routinely collected imaging data for cardiovascular risk assessment. This could help shift medicine from reactive treatment to proactive prevention, ultimately reducing morbidity, mortality, and healthcare costs.
Source: Mass General Brigham
Image Credit: iStock
References:
Hagopian R, Strebel T, Bernatz S et al. (2025) AI Opportunistic Coronary Calcium Screening at Veterans Affairs Hospitals. NEJM AI. 2(6).
