Detection of Atrial Fibrillation From PPG Sensor Data Using Variational Mode Decomposition
Abstract
Atrial fibrillation (AF) is one of the most common arrhythmia. AF can be a reason for strokes and damage to heart activities. The electrocardiogram (ECG) is commonly used for AF detection. It should be noted that the burst AF does not show any symptoms and is difficult to detect using ECG. As an alternative to ECG, photoplethysmography (PPG) is used for AF diagnosis, which is easy to record and suitable for long-term monitoring. This letter proposes a new classification framework for automated AF detection based on variational mode decomposition (VMD). The proposed framework has been studied on a publicly available dataset. The proposed VMD-based classification framework outperformed other state-of-the-art methods used for AF detection and achieved the highest accuracy of 99.08% using ten-fold cross-validation. The results of the proposed methodology have shown significant improvement, and the developed sensor-based system has proved suitability to real-time clinical practices.
