Adaptive filtering for noise reduction in photoplethysmography signals

This article presents an innovative hybrid filtering approach applied to photoplethysmography (PPG) signals, based on the Pulse-sensor, widely used in physiological monitoring devices. The sensor combines hardware functionality, including conventional analogue filtering based on resistor and capacitor net-works, with the ability to access raw signals via a digital interface. To take full advantage of this capability, we designed a soft filtering system fully implemented in the MATLAB development environment. This digital filtering enables real-time processing of photodiode data, bypassing the limitations inherent in fixed analogue filtering. By combining the advantages of both approaches – hardware robustness and software flexibility – our hybrid system improves PPG signal quality, notably by reducing artefacts linked to motion and ambient noise. The performance of this method has been evaluated through a series of test bench experiments, demonstrating a significant improvement in the accuracy of physiological parameter extraction. This re-search opens up new prospects for the development of intelligent, reliable wearable devices for continuous health monitoring in a variety of clinical and ambulatory settings.