Deep learning detects heart failure with preserved ejection fraction using a baseline electrocardiogram
Why the study?
HFpEF diagnosis relies on clinical, invasive, and laboratory tests, and no known typical ECG features currently exist.
Does a deep learning-enabled CNN applied to baseline ECGs accurately detect HFpEF in patients at risk?
Population
1884 patients with exertional dyspnoea and EF ≥50%, plus 203 external validation volunteers
Comparison
CNN classification vs ESC diagnostic criteria for HFpEF
Design
Development and external validation study of a deep learning model
Authors
Discussion
Loading...
Member takes
Implication
May aid noninvasive HFpEF detection in dyspnoea; leaves open prospective validation before clinical adoption.
Structured PICO
Does a deep learning-enabled CNN applied to baseline ECGs accurately detect HFpEF in patients at risk?
P
Population
1884 patients presenting with exertional dyspnoea or equivalent, preserved ejection fraction (≥50%), and clinical suspicion for coronary artery disease, plus an external validation cohort of 203 volunteers in a prospective heart failure screening programme.
I
Intervention
Convolutional neural network (CNN) applied to baseline electrocardiogram (ECG)
C
Comparator
European Society of Cardiology (ESC) criteria for HFpEF (including NT-proBNP measurements)
O
Outcome
Detection of HFpEF (measured by AUC, sensitivity, specificity, PPV, NPV)surrogate
A deep learning algorithm applied to standard baseline ECGs can detect HFpEF with high sensitivity, offering a potential screening tool for patients at risk.
Cite This Study
Unterhuber et al. (2021) studied this question.