Coronary artery calcium (CAC) testing is a method for assessing heart attack risk that remains underused despite its potential benefits. Over time, plaque in the arteries undergoes a transformation from lipid-rich material to calcium deposits. Heart attacks often occur due to the rupture of younger, lipid-rich plaques, which can trigger inflammation and clotting, obstructing blood flow to the heart. While calcified plaques are typically stable, the presence of CAC can indicate that more unstable, rupture-prone plaques may also be present.
CAC is generally identified through dedicated heart-specific CT scans, although it can sometimes be detected on routine chest CTs. The development of algorithms to calculate CAC scores from these standard scans could enhance access to this risk assessment tool. Implementing these algorithms could enable healthcare providers to identify patients with elevated CAC scores, potentially prompting them to seek additional care. Although the presence of startups focused on AI-derived CAC scores is limited, their growth may lead to the identification of high-risk individuals who might otherwise remain undetected.
Traditionally, CAC scans have been regarded as having limited value and are not widely covered by insurance. However, this perspective may be evolving, as more medical experts advocate for the use of CAC scores to refine assessments of cardiovascular risk and encourage patients to consider preventive measures like statin therapy.
The rising use of AI-generated CAC scores aligns with a broader movement toward analyzing extensive medical data to uncover hidden health issues. However, questions remain about their effectiveness in universal screening, particularly in light of a 2022 Danish study that found no improvement in mortality associated with CAC screening. As abnormal CAC scores become more prevalent, concerns about follow-up arise, especially since many healthcare systems may not be equipped to manage these incidental findings effectively. Without standardized procedures, there is a risk that the demands of addressing these findings may outweigh the intended benefits.
Source: https://www.technologyreview.com/2025/10/20/1125336/ai-heart-attack-prediction/