Heart rate variability and sleep monitoring using single-channel PPG signal analysis: techniques and applications.
Abstract
INTRODUCTION Heart Rate Variability (HRV) is the variation in Inter Beat Intervals (IBIs) of a heart signal and a measure of the cardiovascular system's flexibility to changes not expected from other factors. Traditional methods use electrocardiography (ECG) and polysomnography (PSG), which are invasive, costly, and unsuitable for continuous monitoring. Single-channel Photoplethysmography (PPG) is a wearable alternative but is noisy and prone to motion artefacts.
Methods
The proposed Single-Channel PPG-Based Hybrid HRV-Sleep Analysis (SCH-HSA) method applies Variational Mode Decomposition (VMD) to enhance signal quality. Peaks are identified, the Hilbert transform calculates IBIs, and sample entropy extracts nonlinear HRV features. Particle Swarm Optimisation (PSO) retains the most relevant features, while a Bidirectional Long Short-Term Memory (Bi-LSTM) model learns temporal patterns and categorises sleep stages.
Results
Experimental evaluation achieved 94.6% HRV estimation accuracy and 95.2% sleep-stage classification accuracy, outperforming Gudi et al., GOA-IBI, SleepPPG-Net, and SC-PPG for real-time wearable healthcare applications.
Conclusion
Experimental evaluation achieved 94.6% HRV estimation accuracy and 95.2% sleep-stage classification accuracy, outperforming Gudi et al., GOA-IBI, SleepPPG-Net, and SC-PPG for real-time wearable healthcare applications.