Modulus-matched epidermal biointerfaces for robust surface electromyography sensing

Surface electromyography (sEMG) has emerged as a promising modality for intuitive and efficient human–machine interfacing. However, mismatches in Young’s modulus between conventional epidermal electrodes and human skin often lead to poor interfacial contact, resulting in unstable contact impedance and pronounced motion artifacts; this is particularly relevant under dynamic and noisy conditions. To address this issue, we develop a modulus-matched epidermal electrode with a Young’s modulus tuned to 121 kPa in order to match human skin. The electrode shows superior skin conformity and stable contact impedance under skin deformation. Consequently, it delivers lower motion artifacts (28.9 µV) and a higher signal-to-noise ratio (18.4 dB) compared to commercial Ag/AgCl electrodes, even during vibrational disruptions. To boost recognition accuracy, a novel deep learning algorithm is established, which introduces a deconvolutional feature reconstructor to facilitate rapid adaptation to non-ideal factors. Then, synergistically integrating the proposed electrode and algorithm, a hand gesture recognition system is developed for robotic arm control. This system demonstrates a minimal accuracy decline of 1.6 percentage points (pp) under external interference, outperforming the Ag/AgCl-based system, which shows a 22.1 pp drop. This study underscores the importance of modulus matching in electrode design, providing new insights and a reproducible framework for developing bioelectronic interfaces that can perform complex human–machine interactive tasks while withstanding real-world perturbations.