Researchers from Edith Cowan University (ECU) and the University of Manitoba have created an automated program designed to detect cardiovascular issues and fall risks through routine bone density scans.
This innovation could significantly aid in identifying critical health problems before they escalate to life-threatening levels.
The algorithm, developed by ECU’s Dr. Cassandra Smith and Dr. Marc Sim, evaluates vertebral fracture assessment (VFA) images obtained during standard bone density exams frequently used in osteoporosis management.
By examining the presence and severity of abdominal aortic calcification (AAC) in these images, the program can swiftly identify patients at risk for heart attacks, strokes, and severe falls.
Impressively, the algorithm operates at remarkable speed. While an experienced reviewer might take five to six minutes to derive an AAC score from a single scan, the machine learning system can analyze thousands of images and generate predictions in under a minute.
This efficiency could greatly benefit healthcare systems aiming to screen large populations for concealed health threats.
The necessity for such screenings is clear; research by Dr. Smith revealed that an alarming 58% of older adults undergoing routine bone density scans exhibited moderate to high levels of AAC.
Worryingly, one in four individuals with elevated AAC levels had no prior knowledge of their increased risk.
“Women are often under-screened and under-treated for cardiovascular diseases,” Dr. Smith remarked. “This study illustrates that we can leverage widely available, low-radiation bone density equipment to identify women at high risk, enabling them to pursue appropriate treatment.”
Moreover, the algorithm’s capabilities extend beyond heart health. Dr. Sim discovered that patients with moderate to high AAC scores had a significantly higher risk of hospitalization due to falls and fractures compared to those with lower scores.
“The greater the calcification in your arteries, the increased likelihood of falls and fractures,” Dr. Sim noted. While conventional fall risk factors such as previous falls and low bone density are well understood, vascular health is often overlooked.
“Our analysis demonstrated that AAC is a strong predictor of fall risk, even more so than other recognized factors,” he added.
As with any emerging technology, there are challenges to address before AI-assisted screening becomes commonplace in clinical practice.
Foremost, the algorithm must be validated in larger, more diverse populations and seamlessly integrated into current clinical workflows.
If these hurdles are surmountable, a straightforward bone scan—something millions of older adults already get regularly—might transform into an early warning system for some of the most prevalent and serious health concerns we face.
