Automated abdominal aortic calcification scoring and long-term falls and fracture risk: the Perth Longitudinal Study of Ageing Women

Automated abdominal aortic calcification scoring and long-term falls and fracture risk: the Perth Longitudinal Study of Ageing Women

Abadi Kahsu Gebre^1,2, Marc Sim ^1,3, Syed Zulqarnain Gilani^1,4, Naeha Sharif^4,5, David Suter^1,4, Douglas P Kiel⁶, William D Leslie⁷, John T Schousboe⁸, Richard L Prince^1,3, Joshua R Lewis<sup>1,3,9

1</sup>Nutrition & Health Innovation Research Institute, School of Medical and Health Sciences, Edith Cowan University, Joondalup, WA, Australia
²School of Pharmacy, College of Health Sciences, Mekelle University, Mekelle, Tigray, Ethiopia
³Medical School, The University of Western Australia, Perth, WA, Australia
⁴Centre for AI&ML, School of Science, Edith Cowan University, Joondalup, WA, Australia
⁵School of Physics, Mathematics and Computing, The University of Western Australia, Perth, WA, Australia
⁶Hinda and Arthur Marcus Institute for Aging Research, Hebrew Senior Life, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, USA
⁷Departments of Medicine and Radiology,University of Manitoba, Winnipeg, Manitoba, Canada
⁸Park Nicollet Osteoporosis Center and HealthPartners Institute, HealthPartners, Minneapolis, Minnesota, USA
⁹Centre for Kidney Research, Children’s Hospital at Westmead, School of Public Health, Sydney Medical School, The University of Sydney, Sydney, NSW, Australia

Abstract

Abdominal aortic calcification (AAC) is a recognised measure of vascular disease that is an emerging predictor of cardiovascular events. AAC can be detected on lateral spine images used for vertebral fracture assessment at the time of bone density screening. We and others have identified clinically significant associations between AAC, muscle strength decline and falls and fractures. Current AAC assessment requires manual evaluation using a 24-point scoring method (AAC-24). To address this major limitation, an automated machine-learning algorithm (ML-AAC-24) was developed to assess AAC-24 scores from lateral spine scans. This study examined the relationship between ML-AAC24 and (i) injurious falls and ii) fractures risk in older women (mean age, 75 ± 3 years). Over 14.5 years, 413 (39.3%) injurious falls were recorded from linked health records, whilst over 10 years 253 (24.7%) clinical and 68 (6.6%) vertebral fractures that were self-reported and verified from medical records were recorded. Using our algorithm, women with moderate to severe ML-AAC (score=2+) had a greater risk of injurious falls (HR 1.36, 95%CI 1.01-1.67), clinical fractures (HR 1.45, 95%CI 1.12-1.87), and vertebral fractures (HR 1.76, 95%CI 1.04-2.97) compared to those with low ML-AAC (score<2). Notably, the relative hazard estimates obtained through machine learning were similar to those previously reported using manual assessment methods. In conclusion, this novel automated method for assessing AAC is associated with long-term incident injurious falls and fractures. This algorithm can provide further clinically important prognostic information that could be easily captured at the same time as bone density testing for fracture risk. Funding: This study is supported by FHRI Fund Grant (ID WANMA2021/6 Lewis)

Biography

Abadi Gebre is a PhD candidate at Edith Cowan University’s Nutrition and Health Innovation Research Institute. His research focuses on identifying novel cardiovascular causes of falls and fractures to promote healthy aging. Prior to his PhD studies, he spent over five years as a lecturer and researcher oversea. Mr. Gebre has a strong track record of publishing over 25 peer-reviewed articles in reputable journals.