AI Can Now Predict Your Heart Attack Risk From a Simple Scan

A new AI tool called CARDINAL is changing how doctors predict heart attacks and strokes — and it only needs a standard, non-contrast CT scan to do it. That means no extra tests, no invasive procedures, just routine imaging most people already get. Developed by a team of researchers, CARDINAL uses deep learning to analyze subtle patterns in the scan that humans miss, turning those patterns into a personalized 10-year heart risk score.

In a massive study of 17,659 patients, CARDINAL outperformed traditional risk calculators and even advanced imaging techniques. For predicting heart events 10 years in advance, it achieved an accuracy score (AUROC) of 0.866 — significantly better than older methods. It also more accurately separated high-risk from low-risk patients, with a hazard ratio over 10 times higher for those flagged as high risk.

What makes this groundbreaking is that it works with routine CT scans already used for other reasons — like checking for lung issues or chest pain. No extra cost, no extra radiation. For patients, this could mean earlier interventions like diet changes or medications. For doctors, it’s a more reliable way to decide who needs aggressive treatment versus watchful waiting.

The catch? This is still research. While promising, CARDINAL isn’t in clinics yet. But if it scales safely, it could become a game-changer in preventative heart care — making early warnings faster, cheaper, and more accurate than ever before.