Scan-based aging scores added risk information beyond blood tests
Among adults aged 20–60, adding scan-based aging scores to a model with blood markers improved disease risk ranking by 0.048 on the concordance index.
32,882 subjects
Limit: The main subgroup was chosen because its age range showed the strongest associations
Edited by Boris Djordjevic · AgingResearch.News
An observational study analyzed 68,682 abdominal computed tomography (CT) scans from 32,882 people to test scan-based estimates of biological age. Researchers trained models on scans classified as healthy, using the shape and texture of five abdominal organs. They defined an age gap as estimated age minus actual age. Associations with future disease were strongest in younger and middle-aged adults, so the main analyses focused on ages 20–60.
The combined age-gap score improved prediction of disease and death beyond age, sex and body mass index. It also improved disease risk ranking beyond selected blood markers. The concordance index measures how well a model ranks people by risk. Individual organ scores added information beyond the combined score for some diseases, including cardiovascular disease and diseases of the liver and pancreas.
Why it matters
Aging can differ across organs. The work addressed whether estimates based on organ structure could help assess future disease risk beyond a person’s age and blood tests.
Caveats
This observational analysis used age-prediction scores as proxies for biological aging, rather than measuring aging itself. The main risk analyses focused on adults aged 20–60, and the preprint has not been peer-reviewed.
- Machine learning
- Cardiovascular disease
- Hepato-pancreatic disease
- Abdominal CT age
- Biological age gap
- Humans
The paper
Cedars-Sinai Medical Center
medRxiv, May 27, 2026, Open access, Preprint, not peer-reviewed
