AI Detects Hidden Heart Disease Risk in Routine Bone Scans, Unlocking Early Prevention

For decades, the dual-energy X-ray absorptiometry (DEXA) scan has been a single-purpose tool, quietly checking the bone density of older adults to diagnose osteoporosis.[1][5]

But hidden within those routine grayscale images of the lateral spine lies a secondary, life-saving piece of data: the health of the body's largest artery.[2][6]

Now, an international research team led by Australian scientists has developed an artificial intelligence algorithm capable of extracting that hidden data in seconds, effectively turning a standard bone scan into a powerful early warning system for cardiovascular disease.[1][3]

The breakthrough centers on a condition known as abdominal aortic calcification (AAC). As cardiovascular disease develops, calcium deposits begin to build up in the walls of the abdominal aorta.[2][5]

Crucially, this calcification in the abdomen often appears years before similar blockages develop in the coronary arteries near the heart.[2][7]

"Vascular calcification often starts in the abdominal aorta before blood vessels such as the heart," explains Professor Joshua Lewis of Edith Cowan University (ECU), the lead researcher on the project. "If you can detect it earlier, then you can make those changes... to prevent going on to having a heart attack or stroke."[3][4]

Historically, spotting AAC on a DEXA scan has been a frustratingly manual process. Highly trained specialists must visually inspect the images, a meticulous task that takes between five and fifteen minutes per scan.[5][6]

Because of the time and expense involved, this secondary analysis is rarely performed during routine clinical practice, leaving a wealth of preventative health data locked away and unread.[2][6]

The newly developed AI software eliminates this bottleneck entirely. Trained on tens of thousands of medical images, the algorithm can analyze a DEXA scan and quantify the extent of aortic calcification in a fraction of a second.[1][5]

Operating at peak capacity, the software is capable of processing roughly 60,000 images in a single day, achieving an 80% agreement rate with expert human readers in its initial real-world testing phase.[5][6]

The implications for public health screening are massive. In Australia alone, approximately 700,000 people undergo a DEXA bone scan every year.[3][4]

By running those existing scans through the AI algorithm, hundreds of thousands of patients could simultaneously learn about their cardiovascular risk without requiring a separate doctor's appointment, an additional imaging procedure, or any extra radiation exposure.[1][7]

For patients, this invisible, frictionless screening can be life-altering. Pippa Luke, a patient who discovered she had advanced blood vessel disease through the scanning program, described the revelation as a crucial wake-up call.[3][4]

"Getting the results was definitely a catalyst for increasing my knowledge and awareness of heart health because otherwise it's very silent—you don't know you have anything," Luke noted.[3][4]

Armed with the early warning, she was able to proactively overhaul her lifestyle, adopting a Mediterranean diet and a targeted regimen of cardio and resistance exercises to halt the disease's progression before a catastrophic event occurred.[3][4]

Beyond heart attacks and strokes, the presence of AAC is a recognized predictor for a cascade of other late-life health complications, including an elevated risk of falls, severe fractures, and late-life dementia.[5][6]

Identifying high-risk individuals early allows healthcare providers to implement comprehensive, multi-disciplinary care plans that address both cardiovascular and cognitive health simultaneously.[1][6]

The project, supported by the Heart Foundation's Catalyst Partnership program and recently published in the Journal of the American College of Cardiology: Advances, is now moving aggressively toward real-world clinical integration.[2][7]

Researchers are currently seeking additional philanthropic and private funding to refine the algorithm further and deploy it across global healthcare networks.[3][4]

As artificial intelligence continues to reshape modern medicine, this breakthrough highlights a profound shift: the most impactful AI applications may not always involve inventing new medical devices, but rather teaching computers to see the life-saving clues we have been capturing all along.[1][5]