A streamlined cardiovascular-magnetic-resonance (CMR)-derived machine-learning model for estimating cardiovascular biological age: development and validation in the UK-biobank and multi-ethnic study of atherosclerosis
An erratum to this article was published (Eur Heart J Cardiovasc Imaging, 1 Sep 2026 ↗)
Authors' conclusion
A biologically older cardiovascular system was independently associated with adverse cardiovascular outcomes across both sexes. In females, advanced cardiovascular ageing also predicts all-cause mortality, irrespective of chronological age.
Study context
- Population
- UK-Biobank participants (development in healthy participants) with validation in Multi-Ethnic Study of Atherosclerosis (MESA) participantsDevelopment, healthy UK-Biobank participants 3,760 · UK-Biobank association cohort 31,784 · MESA validation 897
- Comparison
- HeartAge-gap (difference between HeartAge and chronological age) per one-year increase, for cardiovascular outcomes and mortality
Methodological detailsgradient-boosting regression, validated in MESA
- Gradient-boosting regression
- Validated in MESA
Performance
As reported in the abstract
Composite:
- Cardiovascular outcome
- All-cause mortality (events)
Composite cardiovascular outcome
2316
7.3%
All-cause mortality
363
1.1%
Further findings
| Outcome | HR (95% CI) | P value | |
|---|---|---|---|
| Composite cardiovascular outcome per one-year increase in HeartAge-gap, UK-Biobank — Each one-year increase in HeartAge-gap · females | 1.022 (1.001–1.044) | P = 0.048 | |
| Composite cardiovascular outcome per one-year increase in HeartAge-gap, UK-Biobank — Each one-year increase in HeartAge-gap · males | 1.017 (1.002–1.033) | P = 0.027 | |
| All-cause mortality per increase in HeartAge-gap, UK-Biobank (females only) · females · regardless of chronological age | 1.061 (1.007–1.118) | P = 0.027 | |
| Cardiovascular outcome per increase in HeartAge-gap, MESA (females only) · female MESA participants · independently of chronological age and other confounders | 1.113 (1.025–1.210) | P = 0.011 | |
| 0.512Hazard ratio (log scale) |
- Composite cardiovascular outcome per one-year increase in HeartAge-gap, UK-Biobank: independently of chronological age and confounders including, body-mass-index, ischaemic heart disease, diabetes, and hypertension
Original publication
A streamlined CMR-derived machine-learning model for estimating cardiovascular biological age: development and validation in the UK-biobank and multi-ethnic study of atherosclerosis. Eur Heart J Cardiovasc Imaging. Published 4 December 2025. doi:10.
Summarised from the published abstract; every number is checked against it