AI-Supported Scans Measuring Heart Fat Could Better Predict Cardiovascular Risk

Why It Matters

The addition of AI‑measured heart fat refines risk stratification, enabling clinicians to target preventive therapies to patients who might otherwise be missed, potentially reducing future heart events and healthcare costs.

Key Takeaways

• AI extracts pericardial fat volume from routine CAC scans automatically.
• Higher pericardial fat predicts cardiovascular events independent of traditional risk factors.
• Adding AI‑derived fat improves risk prediction for low‑ and intermediate‑risk patients.
• No extra imaging needed; measurement uses existing CAC data.
• Study followed ~12,000 adults over 16 years, confirming long‑term predictive value.

Pulse Analysis

Coronary artery calcium (CAC) scans are already a staple in preventive cardiology, offering a quick snapshot of arterial plaque. Recent advances in artificial intelligence now allow those same images to be repurposed for a second, often overlooked metric: pericardial fat volume. This visceral fat depot sits around the heart and has long been linked to inflammation and metabolic disturbances that accelerate atherosclerosis. By training deep‑learning models on manually annotated scans, researchers have created an automated tool that can segment and quantify this fat with high precision, all without requiring any additional imaging or radiation exposure.

In a longitudinal study of roughly 12,000 adults tracked over 16 years, AI‑derived pericardial fat emerged as an independent predictor of cardiovascular events. When the fat measurement was layered onto traditional risk equations such as the AHA PREVENT model and the CAC score itself, the combined model re‑classified a notable share of low‑ and intermediate‑risk individuals into higher‑risk categories. This re‑classification is clinically meaningful because it highlights patients who could benefit from earlier statin therapy, lifestyle counseling, or more intensive monitoring—interventions that might otherwise be delayed under conventional risk assessments.

The practical implications are substantial. Clinicians can now extract richer diagnostic information from an existing scan, improving decision‑making without added cost or patient inconvenience. However, integration into routine workflows will require validation across diverse populations, clear reimbursement pathways, and education for providers on interpreting the new biomarker. As AI continues to mature, pericardial fat measurement may become a standard component of cardiovascular risk profiling, exemplifying how machine‑learning‑enhanced imaging can bridge gaps in preventive care.