A novel hybrid filtering paradigm for drastically reducing noise in photoplethysmography signals: a performance leap beyond traditional approaches

Photoplethysmography (PPG) is a widely used non-invasive technique for continuous cardiovascular monitoring, offering valuable health data such as heart rate and oxygen saturation. However, motion artifacts (MAs) caused by body movements during signal acquisition significantly degrade the quality of PPG signals, leading to inaccurate health monitoring. This study proposes a hybrid denoising method combining wavelet transforms, adaptive filtering, and Kalman filtering to remove motion artifacts in PPG signals effectively. The method decomposes the signal into multiple frequency bands, adjusts for dynamic noise levels, and smooths the signal to improve accuracy. Experimental results show that the proposed method significantly outperforms traditional filtering techniques, achieving a 32.02% reduction in mean squared error (MSE) compared to the adaptive Kalman filter in Gaussian noise simulations, and a 51.46% reduction in MSE in Poisson noise simulations. The proposed method also achieved the highest peak signal-to-noise ratio (PSNR) values, demonstrating superior performance in noise removal while preserving the integrity of the PPG signal. Despite these promising results, the method’s longer processing time remains a limitation for real-time applications, particularly for wearable devices with limited computational power. Future work should focus on optimizing processing speed and testing the method under diverse noise conditions to ensure its practical applicability in real-world health monitoring systems.