Groundbreaking Artificial Intelligence Software to Assess Future Health Risks at the Push of a Button

Groundbreaking Artificial Intelligence Software to Assess Future Health Risks at the Push of a Button

Scientists at Edith Cowan University have engineered a program capable of swiftly analyzing scans of bone density to identify abdominal aortic calcification (AAC), a significant marker for cardiovascular incidents and other health-related concerns. With an 80% agreement rate with specialized analysts, this technology holds the potential to significantly alter the landscape of early disease diagnosis within standard medical procedures.

Rapid identification of cardiovascular risk indicators is now feasible through bone density scans.

Utilizing artificial intelligence, it will soon become possible to foresee our vulnerability to severe health issues in the future merely by initiating a software program.

Abdominal aortic calcification (AAC) is characterized by the accumulation of calcium deposits in the abdominal aorta’s walls. This is an indicator for an elevated risk of cardiovascular events such as heart attacks and strokes.

Moreover, AAC serves as a predictor for the likelihood of experiencing falls, fractures, and late-stage dementia. Notably, standard bone density scanning machines, commonly used to diagnose osteoporosis, are also capable of detecting AAC.

Traditionally, the interpretation of these images has required specialized experts, consuming approximately 5 to 15 minutes per scan.

However, a cross-disciplinary team from Edith Cowan University’s School of Science and School of Medical and Health Sciences have devised software capable of analyzing these scans at an extraordinary pace: an estimated 60,000 images within a single day.

Associate Professor Joshua Lewis, a Heart Foundation Future Leader Fellow, stated that this considerable increase in analytical efficiency is pivotal for facilitating AAC’s broader application in research and aiding individuals in mitigating future health issues.

“The swift acquisition of these images and automated scoring at the time of bone density assessments may herald new methodologies for early detection of cardiovascular diseases and ongoing disease monitoring within regular medical practice,” he commented.

Time-Saving Breakthrough

The findings are a product of an international partnership among multiple institutions including Edith Cowan University, the University of Western Australia, the University of Minnesota, Southampton, the University of Manitoba, the Marcus Institute for Aging Research, and Hebrew SeniorLife Harvard Medical School, making it a truly global and interdisciplinary endeavor.

While not the first algorithm tailored to assess AAC from these images, this study stands as the largest of its type, grounded on widely-used bone density machine models and tested in a real-world environment with routine bone density scans. More than 5,000 images were evaluated by both experts and the developed software.

After comparative analysis, both the software and experts concurred on the AAC levels (low, moderate, or high) 80% of the time. Remarkably, only 3% of individuals assessed to have high AAC levels were inaccurately diagnosed as low-level by the software.

“This is of significance as these are the patients with the most advanced stage of the disease and the highest likelihood of life-threatening and non-life-threatening cardiovascular incidents and overall mortality,” stated Professor Lewis.

“Although refinements to the software’s precision in relation to human interpretations remain to be made, these outcomes are derived from our inaugural algorithm, and subsequent versions have already demonstrated enhanced performance.”

“The ability for automated evaluation of AAC presence and its extent, with accuracy comparable to imaging specialists, opens the door for extensive screening for cardiovascular diseases and other health conditions even before any symptoms manifest,” he added.

This offers individuals at risk an opportunity to undertake requisite lifestyle modifications at an earlier stage, thereby better positioning them for improved health in their later years.

Reference: The cited study is titled “Machine learning for abdominal aortic calcification assessment from bone density machine-derived lateral spine images” and has been published in eBioMedicine.
DOI: 10.1016/j.ebiom.2023.104676

Funding for the research was made possible by the Heart Foundation, courtesy of Professor Lewis’ 2019 Future Leadership Fellowship that provided financial support for a span of three years.

Frequently Asked Questions (FAQs) about Artificial Intelligence in Early Disease Detection

More about Artificial Intelligence in Early Disease Detection

• Edith Cowan University Research Programs
• Artificial Intelligence in Healthcare
• The Heart Foundation
• eBioMedicine Journal
• Cardiovascular Disease: Causes, Symptoms, and Treatments
• Machine Learning in Medical Diagnosis
• Bone Density Testing
• Abdominal Aortic Calcification
• DOI Resolver