Physiological signal processing is crucial for next-generation human-machine interfaces. This paper presents a highly efficient neuromorphic physiological signal processing system using VO2 memristors. The system leverages the unique properties of VO2 memristors for asynchronous spike encoding and constructing Leaky Integrate and Fire (LIF) and Adaptive-LIF (ALIF) neurons within a Long short-term memory Spiking Neural Network (LSNN). The system achieves high accuracies (95.83% for arrhythmia classification and 99.79% for epileptic seizure detection) using small LSNNs, highlighting the potential of memristors in efficient neuromorphic systems for human-machine interfaces.

Rui Yuan, Pek Jun Tiw, Lei Cai, Zhiyu Yang, Chang Liu, Teng Zhang, Chen Ge, Ru Huang, Yuchao Yang