Robust Photoplethysmography Signal Denoising via Mamba Networks

The paper introduces DPNet, a Mamba-based deep learning framework for denoising photoplethysmography (PPG) signals while preserving physiological information. It incorporates a scale‑invariant signal‑to‑distortion ratio loss and an auxiliary heart‑rate predictor to enhance waveform fidelity and maintain heart‑rate accuracy. Experiments on the BIDMC dataset show that DPNet outperforms conventional filtering and existing neural models in robustness against synthetic noise and real‑world motion artifacts, making it suitable for wearable healthcare systems.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.