Machine learning derived abdominal aortic calcification is associated with physical frailty in community-dwelling

Abadi K Gebre1, Marion Mundt2,3, Elsa Dent4

  • 1Nutrition & Health Innovation Research Institute, School of Medical and Health Sciences, Edith Cowan University, Perth, WA, Australia.

Geroscience
|June 23, 2026
PubMed

Insights

Subclinical cardiovascular disease, measured as abdominal aortic calcification (AAC), is linked to physical frailty. Early detection of frailty may be possible using AAC from routine bone density scans.

Area of Science:

  • Gerontology
  • Cardiovascular Disease
  • Biomedical Imaging

Background:

  • Clinical cardiovascular disease (CVD) is common in frail individuals.
  • The association between subclinical CVD markers like abdominal aortic calcification (AAC) and frailty is not well understood.
  • Frailty is a significant geriatric syndrome associated with adverse health outcomes.

Purpose of the Study:

  • To investigate the cross-sectional relationship between abdominal aortic calcification (AAC) and physical frailty.
  • To assess if a machine learning model for AAC (ML-AAC24) can identify individuals at higher risk of frailty.
  • To explore the potential of AAC as an early marker for frailty detection.

Main Methods:

  • Utilized data from 49,081 participants in the UK Biobank Imaging Study without a prior atherosclerotic CVD diagnosis.
  • Quantified abdominal aortic calcification (AAC) using a validated machine learning model (ML-AAC24) and categorized it into low, moderate, and high severity.
  • Assessed physical frailty using a modified Fried's phenotype, including grip strength, walking speed, weight loss, exhaustion, and physical inactivity.
  • Employed multivariable-adjusted multinomial logistic regression to analyze associations between ML-AAC24 extent and frailty status (pre-frail, frail).

Main Results:

  • Approximately 20% of participants exhibited moderate or high levels of ML-AAC24.
  • Individuals with moderate and high ML-AAC24 showed significantly greater odds of being pre-frail (ORs 1.06-1.14) and frail (ORs 1.27-1.58) compared to those with low ML-AAC24.
  • Associations between ML-AAC24 and frailty status remained significant after adjusting for multiple covariates.
  • Similar associations were observed when stratified by sex.

Conclusions:

  • Higher degrees of abdominal aortic calcification (AAC), as assessed by ML-AAC24, are significantly associated with increased odds of physical frailty.
  • AAC, detectable from lateral spine images during routine bone density testing, may serve as a valuable, non-invasive biomarker for early frailty detection.
  • These findings underscore the importance of considering subclinical CVD markers in comprehensive geriatric care and frailty assessment.