Why the study?
Diagnosis of HFpEF using the HFA-PEFF and H2FPEF scores remains clinically challenging and relies on echocardiographic assessment.
Does automated deep learning interpretation of echocardiograms perform similarly to manual measurements in diagnosing HFpEF?
Population
102 test HFpEF patients, 129 ambulatory HFpEF patients, and 427 diagnostic validation patients (182 HFpEF, 245 non-HFpEF)
Comparison
Automated deep learning echocardiogram interpretation vs manual measurements for HFA-PEFF and H2FPEF scores
Design
Cohort validation study
Key result
Automated deep learning interpretation of echocardiograms yielded similar diagnostic accuracy to manual analysis for HFpEF (AUC for HFA-PEFF 0.70 vs 0.71; AUC for H2FPEF 0.78 vs 0.75).
Authors
Loading...
Automated scoring may match manual HFpEF diagnostic accuracy; extends deep learning echo validation but leaves prospective adoption open.
Observational (n=658)
Does automated deep learning interpretation of echocardiograms perform similarly to manual measurements in diagnosing HFpEF?
Effect estimate: AUC 0.70 (HFA-PEFF) and 0.78 (H2FPEF) (95% CI 0.66-0.74)
Automated deep learning analysis of echocardiograms provides similar diagnostic accuracy to manual measurements for calculating HFA-PEFF and H2FPEF scores in patients with suspected HFpEF.
Venema et al. (2026) conducted an observational in Heart Failure with Preserved Ejection Fraction (HFpEF) (n=658). Automated deep learning interpretation of echocardiograms vs. Manual echocardiographic measurements was evaluated on Diagnostic accuracy (area-under-the-curve) for HFA-PEFF and H2FPEF scores (AUC 0.70 (HFA-PEFF) and 0.78 (H2FPEF), 95% CI 0.66-0.74). Automated deep learning interpretation of echocardiograms yielded similar diagnostic accuracy to manual analysis for HFpEF (AUC for HFA-PEFF 0.70 vs 0.71; AUC for H2FPEF 0.78 vs 0.75).