Body signals such as muscle activity and heartbeats carry valuable information for wearable health monitoring, rehabilitation, and human-machine interaction. However, these electrical signals are very weak and can be easily disturbed by electromagnetic interference from everyday electronic devices, communication systems, and surrounding power sources. This interference can distort signal waveforms, lower signal quality, and reduce the reliability of wearable bioelectronic systems.
A research team from Zhengzhou University has developed a skin-worn Janus bioelectrode designed to keep these signals clear in complex electronic environments. The study, titled An EMI-suppressed, high-fidelity Janus bioelectrode with gradient impedance for accurate dual-biosignal-based motion recognition , was published in Science Bulletin .
The new electrode uses an asymmetric Janus structure. One side is a magnetic-dielectric aerogel made from carbon nanotubes, NiFe2O4, and thermoplastic polyurethane. The other side is a highly conductive liquid-metal textile. Together, these two layers create a gradient impedance structure from the skin interface to the surrounding environment.
This design allows incoming electromagnetic waves to enter the aerogel layer, where they are first weakened by dielectric and magnetic losses. The remaining waves are then reflected by the liquid-metal textile and forced back into the aerogel for further absorption. In this way, the electrode does not simply reflect interference away. Instead, it guides, absorbs, and dissipates electromagnetic noise through an absorption-reflection-reabsorption process.
To test its practical performance, the researchers recorded electromyography signals under strong 50 Hz electromagnetic interference. A commercial Ag/AgCl electrode showed obvious waveform distortion under interference, while the Janus bioelectrode maintained a clearer and more stable signal. Under strong interference, the electrode achieved a signal-to-noise ratio of 16.6 dB.
The team then combined the two types of biosignals with a machine-learning model for human motion recognition. Using only electrocardiography signals, the recognition accuracy was 50.13%. Using only electromyography signals, the accuracy increased to 86.45%. When the two signals were fused, the recognition accuracy reached 97.7% for four representative motion states: sitting, standing, walking, and squatting.
This work offers a signal-integrity-centered strategy for wearable bioelectronics. By combining electromagnetic field regulation, stable skin-electrode coupling, and dual-biosignal machine learning, the Janus bioelectrode provides a promising platform for wearable health monitoring, rehabilitation assessment, intelligent prosthetics, and human-machine interfaces in electromagnetically noisy environments.
The first authors of this paper are Qizhi Yao and Wenke Yang from Zhengzhou University. The corresponding author is Professor Hu Liu. The work was supported by the National Natural Science Foundation of China, the Excellent Youth Fund of the Natural Science Foundation of Henan Province, and the Central Plains Youth Top-notch Talent Program of Henan Province.
Science Bulletin
Experimental study