AI-Optimized Single-Feature Recognition Model for Heart Failure

This prospective, single-center and observational study aims to develop and validate the single-feature artificial intelligence algorithm based on data collected via the wearable ECG patches in patients with heart failure (HF). The main question: Does the algorithm, using synchronized ECG and accelerometer signals from the ECG patches, achieve accurate detection of heart sounds (S1, S2, and in some patients S3, S4) compared with the Eko CORE 500 digital stethoscope in patients with acute exacerbation of HF? It aims to answer: Participants with confirmed HF (NYHA classification II-IV) will first undergo a 2-minute session of simultaneous ECG patches and digital stethoscope recordings, followed by standard 12-lead ECG, and then the repeated ECG patches and 2-minute heart sound recording session. Data will be used for algorithm training and validation. The primary endpoint is the accuracy of heart sound detection via the Vivalink ECG patches compared with the Eko CORE 500 digital stethoscope.