Core–Shell MWCNTs/TPU Fibers Enable Regionally Decoupled Sensing for Imperceptible Sleep Posture Monitoring and Gesture Decoupling

Abstract

Wearable flexible sensing is advancing toward lightweight, integrated multimodal systems, where fiber–based conductive materials exhibit excellent flexibility, weavability, and biocompatibility. However, single–structure sensing units face narrow response dimensions, hard–to–resolve overlapping deformation signals, and incomplete application loops. Herein, core–shell structured MWCNTs/TPU composite sensing (CTS) fibers are fabricated via coaxial wet spinning using MWCNTs as the conductive phase and TPU as the matrix. These fibers exhibit excellent mechanical stability and high sensitivity: they retain conductive networks after 1600 tensile cycles, achieve a gauge factor of 1878.71 at 250–350% strain, and reach a conductivity of 9.11 S·m^–1. Meanwhile, three distinct sensing elements—plain weave (PW), four–strand diagonal braided structure (FDBS), and four-strand same–mat wrapped structure (FSWS)—were woven from the CTS fibers and serially connected to form a multistructure cooperative sensing unit. Integrating the sensing unit into a pillowcase enables imperceptible signal acquisition of sleep posture resistance variation, followed by wireless transmission to the MultiSense APP. Based on the exported data, the dynamic time warping (DTW) algorithm is adopted to accurately recognize and count left–right lateral lying postures. Moreover, attaching the unit to the palm collects hybrid gesture signals; benefiting from the differential deformation response of the three fabric structures, the proposed Inception-BiLSTM–based IBA–PDN model quantifies the relative contribution of nine signal feature components corresponding to distinct structure–deformation combinations and identifies the dominant deformation mode from overlapping hybrid waveforms. This work establishes a complete technical framework from material preparation to terminal application, showing great potential in sleep monitoring and human–computer interaction.