Asymmetric Heterostructured Breathable Janus Nanofibrous Membranes for Deep Learning-Assisted Wearable Sensing.

Long-term skin-interfaced wearable sensing requires integrated sweat management, breathability, thermal comfort, mechanical compliance, and stable signal acquisition, yet reconciling these requirements remains challenging in conventional homogeneous substrates or mechanically stacked multilayer sensors. Herein, we report an asymmetric heterostructured breathable Janus nanofibrous membrane (AHB-JNM) fabricated by in situ layered electrospinning for deep learning-assisted wearable sensing. The membrane combines a compact, low-surface-energy thermoplastic polyurethane/polydimethylsiloxane (TPU/PDMS) fibrous layer for dry skin contact with a porous, hydrophilic thermoplastic polyurethane/polyethylene glycol (TPU/PEG) fibrous layer for moisture transport and conductive network support. This asymmetric architecture promotes directional liquid transport, suppresses reverse liquid penetration, and preserves interconnected channels for air and water-vapor transport. After printing a hybrid multiwalled carbon nanotube/silver nanowire (MWCNT/AgNW) conductive network onto the TPU/PEG layer, the sensor achieves a gauge factor of 121.29, a strain range up to 200%, a rapid response time of 81.2 ms, and stable sensing over 10 000 loading-unloading cycles. The membrane also provides cooling and warming effects of 1.4°C and 4.3°C, respectively. Combined with a deep learning algorithm, the sensor achieves 98% accuracy in Morse code recognition. These results demonstrate a structurally programmed strategy for integrating moisture regulation, thermal management, and stable electromechanical sensing in skin-interfaced wearable systems.