AI-enabled coronary plaque quantification outperforms traditional risk scores
Artificial intelligence (AI)-guided atherosclerosis quantification significantly improves cardiovascular risk prediction for heart attacks, according to new data from the CONFIRM2 Registry presented at the American Heart Association Scientific Sessions 2025 conference.
The study's authors followed 6,550 patients (51.6% female, with mean age of 59) for 4.4 years. The patients all underwent coronary computed tomography angiography (CCTA) with AI-based quantitative coronary CT analysis (AI-QCT) across numerous international sites. The data show that AI-QCT improves the prediction of major cardiovascular events from 62% to 75%.
"This is another step towards earlier identification and treatment for patients," principal investigator Ibrahim Danad, MD, PhD, of Radboud University Medical Center in Nijmegen, The Netherlands, said in a statement. "In this case, it challenges what we know about obstructive disease and further validates the abilities of AI-QCT. This is another fascinating learning we’ve gained from the CONFIRM2 Registry."
The researchers looked at how the addition of AI-QCT would improve risk predictions when added to the European Society of Cardiology Risk Factor-weighted Clinical Likelihood (ESC RF-CL) scores. The area under the ROC curve (AUC) for predicting major cardiovascular adverse events increased from 0.62 to 0.75. The AUC for predicting death/MI increased from 0.60 to 0.71.
CCTA is in both European and U.S. guidelines to evaluate chest pain patients because of its ability to show vessel stenosis and occluded vessels, helping rule out heart attacks quickly. The CCTA images also show different types of plaque, including soft plaques that cause heart attacks, but quantification and rough assessments of the types of plaques were just estimated int he past because it was too time consuming to measure correctly. AI automation, however, has been a game-changer for this technology and opened up new opportunities to improve patient care.
The AI in this analysis was able to identify patients at risk. Nearly 79% of patients presenting with chest pain were classified as having a very low (35.7%) or low (42.7%) likelihood for obstructive coronary artery disease (CAD). Of those with disease, true obstruction rates were only 5.8% and 14.2%, respectively. AI-guided quantification of non-obstructive atherosclerosis successfully stratified these patients for future cardiovascular events.
Despite the fact that so many symptomatic patients had a very low or low likelihood of obstructive disease, Danad said the way AI-QCT can risk-stratify these patients for future events addresses an important clinical gap.
The CONFIRM2 Registry is sponsored by Cleerly to track the results of its AI-QCT software in the real world.
Dave Fornell has covered healthcare for more than 17 years, with a focus in cardiology and radiology. Fornell is a 5-time winner of a Jesse H. Neal Award, the most prestigious editorial honors in the field of specialized journalism. The wins included best technical content, best use of social media and best COVID-19 coverage. Fornell was also a three-time Neal finalist for best range of work by a single author. He produces more than 100 editorial videos each year, most of them interviews with key opinion leaders in medicine. He also writes technical articles, covers key trends, conducts video hospital site visits, and is very involved with social media. E-mail: [email protected]