Heart failure & cardiomyopathies
Heart Failure with Preserved Ejection Fraction Phenogroup Classification Using Machine Learning
Authors' conclusion
ML could successfully stratify Japanese HFpEF patients into three phenogroups (atherosclerosis and chronic kidney disease, atrial fibrillation, and younger and left ventricular hypertrophy groups).
Study context
- Population
- Japanese patients hospitalised for acute decompensated heart failure (HF) with HFpEF (LVEF >50%), Nara Registry and Analyses for Heart Failure (NARA-HF) registry230 validation cohort
- Participants
- 365 patients
- Exposure
- Machine learning (ML)-derived HFpEF phenogroups 1 vs 2 vs 3
- Phenogroup 1n = 125
- Phenogroup 2n = 200
- Phenogroup 3n = 40
- Outcome
- All-cause death and HF readmission
- Follow-up
- Within 5 years
Methodological detailsunsupervised machine learning (variational Bayesian-Gaussian mixture model (VBGMM)), derivation and validation cohorts, hierarchical clustering, supervised ML
- Unsupervised machine learning (VBGMM)
- Derivation and validation cohorts
- Hierarchical clustering
- Supervised ML
Main finding
As reported in the abstract
P = 0.0036
Observational data: associations, not effects.
Further findings
Phenogroup characteristics
- Phenogroup 1 mean age (years)
- 78.9 ± 9.1
- Phenogroup 1 male
- 57.6%
- Phenogroup 1 mean eGFR (mL/min/1.73 m²)
- 28.5 ± 9.7
- Phenogroup 2 mean age (years)
- 78.8 ± 9.7
- Phenogroup 2 body mass index (BMI)
- 22.78 ± 3.94
- Phenogroup 2 women
- 57.5%
- Phenogroup 2 AF
- 56.5%
- Phenogroup 3 mean age
- 63.5 ± 11.2
- Phenogroup 3 BMI
- 27.46 ± 5.85
Reproducibility of phenogroups
- Hierarchical and supervised clustering showed reproducibility of the three phenogroups
Original publication
Heart Failure with Preserved Ejection Fraction Phenogroup Classification Using Machine Learning. ESC Heart Fail. Published 12 April 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →