Preventing heart disease with AI enhanced scans

Associate Professor James Otton, University of New South Wales

Vanguard Grant

Years funded: 2026 - 2027

Heart disease remains the leading cause of death in Australia, yet our ability to predict who will develop heart problems is still inadequate. Many heart attacks occur in people considered "low risk" by current risk calculators. This is particularly true for women, where our existing tools perform poorly.

Our project uses artificial intelligence (AI) to extract previously hidden information from routine coronary calcium CT scans - a simple, low-cost test already used to detect calcium buildup in heart arteries. These scans contain valuable data about heart chamber size, heart muscle thickness, and fat around the heart that current assessment methods ignore, but that strongly predict future heart problems.

We've developed innovative AI software that can measure these additional heart features from standard calcium scans in just two seconds, without requiring patients to undergo additional testing or radiation exposure. Our preliminary results show that these AI measurements are as accurate as specialized heart imaging techniques like MRI.

Using the Multi-Ethnic Study of Atherosclerosis (MESA) database - which contains thousands of calcium scans, heart MRIs, and 15+ years of health outcomes - we will:

1. Validate our AI measurements against gold-standard heart MRI data 2. Develop a more comprehensive heart risk calculator that combines traditional risk factors (like blood pressure and cholesterol) with our AI measurements 3. Test how much better our new risk calculator performs compared to current methods

Our approach has several important advantages: - It extracts additional valuable information from scans people already receive - It requires no extra cost, time, or radiation exposure for patients - It can identify people at risk for various heart conditions beyond just heart attacks - It works with existing clinical workflows and can be easily integrated into current healthcare systems

If successful, our easily applied tool could significantly improve heart disease risk assessment, particularly for women, rural communities, the developing world and others poorly served by current methods. It would help doctors better determine who needs preventive treatments and monitoring, potentially saving many lives.

Our team includes cardiologists, AI experts, and statisticians with experience successfully deploying medical software. We're committed to making this technology widely available and have developed a clear pathway for regulatory approval and clinical implementation. We plan to make our risk calculator freely available to healthcare providers ensuring widespread impact and equity of access..