Heart failure & cardiomyopathies
Development and Validation of Algorithms to Predict Left Ventricular Ejection Fraction Class from Healthcare Claims Data
Authors' conclusion
The algorithms identify reduced or preserved LVEF in HF patients within a nationwide healthcare claims database with high positive predictive value and low rates of false positives.
Study context
- Population
- Heart failure (HF) patients in the French national claims database (SNDS) and a French HF registry
- Comparison
- Claims-based algorithms vs recorded left ventricular ejection fraction (LVEF)
- Outcome
- Algorithm performance for LVEF class
Methodological detailsmachine learning (random forest), cross-validation
- Machine learning (random forest)
- Cross-validation
Performance
As reported in the abstract
| Test or model | AUC | Identified in validation dataset (reinforced algorithm) | PPV | Specificity |
|---|---|---|---|---|
| LVEF ≤ 40% | 0.84 | 50% | 0.88 | 0.96 |
| LVEF < 50% | 0.79 | |||
| LVEF ≥50% | 0.79 |
AUC
LVEF ≤ 40%0.84
LVEF < 50%0.79
LVEF ≥50%0.79
Identified in validation dataset (reinforced algorithm)
LVEF ≤ 40%50%
LVEF < 50%— not stated in the abstract
LVEF ≥50%— not stated in the abstract
PPV
LVEF ≤ 40%0.88
LVEF < 50%— not stated in the abstract
LVEF ≥50%— not stated in the abstract
Specificity
LVEF ≤ 40%0.96
LVEF < 50%— not stated in the abstract
LVEF ≥50%— not stated in the abstract
Further findings
Most important predictive variables
- Delivery of HF medication, sex, age, hospitalization and natriuretic peptide testing, with differing importance by LVEF category
Original publication
Development and Validation of Algorithms to Predict Left Ventricular Ejection Fraction Class from Healthcare Claims Data. ESC Heart Fail. Published 4 March 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →