Multimodal Wearable Sensing for Biomechanics and Biomolecules Enabled by the M-MPM/VCFs@Ag Interface with Machine Learning Pipeline

The addition sensing device of sweat to wearable biostress sensors would eliminate the need for using multiple gadgets for healthcare analysis. Due to the distinct package fashion of sensor interface for biostress and biomolecule, achieving permeability and multifunctionality in an integrated wearable sensor remains a formidable challenge. Here, a viscose fabrics (VCFs)@silver (Ag) sensing material and encapsulating strategy by multi-microporous membranes (M-MPM) are developed, and simultaneous detection for muscle strain and molecule biomarkers in sweat (glucose, lactate, uric acid) is realized. The package interface exhibits good breathability, directional, and fast sweat transport without interception due to its wetting gradient enabled by the 3D-stacking for VCFs@Ag and M-MPM, which subjoins sensing function of sweat on the biostress sensing platform. And, the filtration effect of M-MPM can resist the pollutant interference to the hand-held surface-enhanced Raman scattering spectrum on skin surface. Then, a bidirectional memory network is constructed to correct for the changes in electrical conductivity of VCFs@Ag under the influence of sweat infiltration. With the help of a machine learning pipeline, the accuracy of multimodal recognition increased to 88.6%. As proof of concept, the package interface of M-MPM/VCFs@Ag provides the feasibility of simultaneous monitoring of the muscle strain and sweat with a single sensor.