Deep learning detects heart failure with preserved ejection fraction using a baseline electrocardiogram.

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Diagnostics

Background:

• Heart failure with preserved ejection fraction (HFpEF) is a growing global health concern.
• Current HFpEF diagnosis relies on clinical, invasive, and laboratory assessments, lacking specific ECG markers.
• Electrocardiographic findings in HFpEF are variable, complicating diagnosis.

Purpose of the Study:

• To develop and validate a deep learning model for identifying HFpEF using electrocardiograms (ECGs).
• To assess the performance of a convolutional neural network (CNN) in classifying HFpEF based on European Society of Cardiology (ESC) criteria.
• To integrate NT-proBNP measurements into the diagnostic algorithm for HFpEF screening.

Main Methods:

• A convolutional neural network (CNN) was trained on ECG data from 1884 patients in a derivation cohort.
• The CNN was designed to classify patients as having HFpEF or being controls, according to ESC criteria.
• The CNN's performance was validated on an external cohort of 203 volunteers in a prospective screening program.

Main Results:

• The CNN achieved an area under the curve of 0.80 for detecting HFpEF in the external validation cohort.
• The model demonstrated high sensitivity (0.99) and negative predictive value (0.98) for HFpEF detection.
• Specificities and positive predictive values were 0.60 and 0.68, respectively.

Conclusions:

• This study presents the first deep learning-enabled CNN for identifying HFpEF according to ESC criteria.
• The CNN demonstrated convincing screening performance when validated on an external cohort of at-risk patients.
• The model shows potential for improving the diagnostic process for HFpEF in at-risk populations.