Validating a Machine Learning Algorithm Using Photoplethysmography for Early Detection of Atrial Fibrillation in Heart Failure Patients During Remote Monitoring

What this Trial Is About

Researchers are evaluating a new machine-learning algorithm designed to detect atrial fibrillation AF from photoplethysmography PPG signals during remote monitoring. The study focuses on patients with heart failure who may have permanent or paroxysmal AF, aiming to validate the algorithms ability to identify clinically relevant AF episodes lasting 30 seconds or longer. This prospective validation study involves comparing the algorithms performance against the gold-standard 12-lead ECG recordings to improve early detection and intervention for AF in heart failure patients. The algorithm uses advanced signal processing of PPG waveforms to assess pulse variability and rhythm irregularities associated with AF. Validation occurs in three stages internal cross-validation, external validation using an independent cohort with paired PPG and ECG recordings, and testing in patients with paroxysmal AF and frequent rhythm changes. The study enrolls approximately 1,000 unique PPG recordings and integrates the algorithm into the Seerlinq remote monitoring platform, which is based on technology from the CE-certified HeartCore device. Participants undergo 12-lead ECG testing to confirm heart rhythm classification, providing a reference standard for evaluating the algorithms diagnostic accuracy. Researchers measure the algorithms performance through various metrics such as sensitivity, specificity, positive and negative predictive values, and classification accuracy. The study is ongoing with data collection and analysis planned through November 2026, focusing on assessing the algorithms ability to detect AF early in heart failure patients during remote monitoring.

A Photoplethysmography-Based Machine Learning Algorithm for Early Atrial Fibrillation Detection: A Prospective Validation Study