Contrastive Machine Learning to Quantify Hypertensive Multiorgan Damage and Identify New Disease Phenotypes: A Multinational Multimodal Study
Authors' conclusion
Machine learning-derived global organ damage scores are feasible in hypertension and enable identification of distinct hypertension-associated organ-disease phenotypes.
Study context
- Population
- UK Biobank imaging substudy participants; external validation in ARICARIC validation n = 5507
- Participants
- 27,099 participants
- Outcome
- HyperScore (global organ damage score)
Methodological detailsmultinational, multimodal, semisupervised contrastive trajectory inference, internal and external validation
- Multinational
- Multimodal
- Semisupervised contrastive trajectory inference
- Internal and external validation
Performance
As reported in the abstract
- AUC for identifying severe end-organ disease
- 0.964 (0.941-0.987)
- Cross-validation mean root mean square error
- 0.104±0.084
- Survival odds differed across HyperScore stages; stratification by BP nonsignificant
- P<0.001
Further findings
External testing in ARIC
- Jensen-Shannon distances for HyperScore distributions as low as
- 0.10
- No significant deviation in organ damage progression patterns
- P > 0.05
Original publication
Contrastive Machine Learning to Quantify Hypertensive Multiorgan Damage and Identify New Disease Phenotypes: A Multinational Multimodal Study. Circulation. Published 21 June 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →