Heart failure & cardiomyopathiesDiagnostic or prognostic study
Deep Phenotyping of Heart Failure with Preserved Ejection Fraction Through Multi-Omics Integration
- Diagnostic accuracy study
- machine learning
- multi-omics
- hold-out validation
Authors' conclusion
We identified HFpEF phenotypes at an early stage, often several years before the onset of clinical symptoms, when the disease trajectory may still be amenable to modification.
Study context
- Population
- UK Biobank participants; subsequent symptom-defined heart failure with preserved ejection fraction (HFpEF) during longitudinal follow-uptraining subset
- Participants
- 401,917 participants
- Outcome
- Detection of subsequent symptom-defined HFpEF
Performance
As reported in the abstract
Detection of subsequent HFpEF in hold-out validation subset
ROC AUC
0.931 (95% CI 0.930-0.931)
Sensitivity
0.857 (95% CI 0.855-0.860)
Specificity
0.847 (95% CI 0.846-0.847)
Further findings
Lead time before symptom onset in asymptomatic individuals who developed HFpEF
- Average years before symptom onset (± 3.9)
- 6.3
Discrimination of high-risk similarity network fusion (SNF) cluster (elevated mortality, dysregulated inflammatory pathways)
- ROC AUC
- 0.988 (95% CI 0.985-0.990)
Original publication
Deep Phenotyping of Heart Failure with Preserved Ejection Fraction Through Multi-Omics Integration. Eur J Heart Fail. Published 22 September 2025. doi:10.1002/ejhf.70041
Summarised from the published abstract; every number is checked against it