AI Tools Aim to Catch Heart Disease in Routine Chest Scans, But Few Hospitals Deploy Them

Artificial intelligence algorithms can identify early signs of heart disease in chest CT scans that patients already receive for other reasons, but uptake among U.S. health systems remains low despite FDA authorization of multiple tools. An estimated 20 to 40 percent of incidental coronary artery calcium — a marker of heart attack and stroke risk — goes unreported annually across 19 million general chest scans performed each year.

Radiologists performing chest CT scans for suspected lung cancer or respiratory issues frequently spot coronary artery calcium as a side finding. The presence and extent of this calcium correlates directly with cardiovascular risk, yet many cases slip through undetected. Nish Khandwala, CEO of Bunkerhill, which has received FDA authorization for its calcium-detection algorithm, described the potential: "Without anybody needing to lift a finger on a day-to-day basis, patients can get screened for cardiac diseases."

Multiple companies have developed FDA-authorized AI systems capable of flagging this incidental calcium in existing scans. The technology requires no additional imaging, no extra cost per scan, and no radiologist intervention beyond reviewing the algorithm's findings. Yet adoption across hospital systems and imaging centers has stalled.

The barriers to deployment are practical and economic. Health systems must integrate new software into existing workflows, train radiologists on the tools, and determine who pays for the screening service when it was not originally ordered. Insurance coverage for opportunistic screening remains inconsistent, leaving many facilities without financial incentive to implement the systems.

THE DETAILS section continues: Katie Palmer's reporting for STAT identified workflow integration and reimbursement as the primary obstacles preventing wider adoption, despite the clinical potential to identify thousands of at-risk patients annually who would otherwise remain unaware of their cardiovascular risk.

Context

Coronary artery calcium scoring has been studied as a cardiovascular risk predictor for decades. The 2019 American College of Cardiology guidelines acknowledged that incidental calcium findings in CT scans can inform treatment decisions, yet no standardized protocol exists for systematic review across U.S. imaging centers. Prior studies have shown that patients informed of incidental calcium findings are more likely to pursue preventive medications like statins, yet the fragmented nature of American imaging means many findings never reach patients or their physicians.

The missed-screening problem extends beyond calcium detection. A 2023 analysis found that incidental findings in imaging studies — ranging from nodules to structural abnormalities — frequently go unreported or fail to reach the ordering physician, representing a systemic gap in patient care that extends across multiple organ systems and imaging modalities.

CONTEXT continues: The financial model for opportunistic screening remains unresolved. Unlike screening mammography or colonoscopy, which have established billing codes and insurance coverage, incidental calcium screening lacks a defined reimbursement pathway. Health systems deploying AI tools must absorb implementation costs without guaranteed revenue recovery, creating institutional resistance despite clinical evidence of benefit.

What's Next

The adoption trajectory for these AI tools will depend on reimbursement policy decisions by major insurers and Medicare. If coverage expands for incidental calcium screening, health systems have the technical infrastructure to deploy algorithms rapidly across existing imaging workflows. Conversely, if reimbursement remains fragmented, adoption will likely remain confined to academic medical centers and large health systems with capital for unfunded initiatives.

Regulatory bodies including the American College of Radiology are developing standardized reporting protocols for incidental calcium findings. These guidelines, expected to be finalized within the next 12 to 18 months, may establish the clinical legitimacy needed to drive insurance coverage decisions and institutional adoption of AI screening tools.