2H-MoS2 lubrication-enhanced MWCNT nanocomposite for subtle bio-motion piezoresistive detection with deep learning integration☆

Abstract

Intelligent piezoresistive health monitoring systems integrate advanced nanocomposite architectures with precise algorithmic analysis for real-time physiological assessment. However, existing works often prioritize high sensitivity at the expense of strain tolerance and require complex fabrication procedures. Herein, we present an environmentally friendly, low-cost, and nonionic fabrication approach for a 2H-phase molybdenum disulfide (2H-MoS2)-enhanced multi-walled carbon nanotube (MWCNT) strain sensor, developed via a systematically optimized vacuum-assisted filtration process. This study is the first to validate the dual enhancement effect of MoS2, leveraging its shear-exfoliation properties to simultaneously improve strain gauge performance and mechanical robustness. The resulting nacre-like layered hybrid nanocomposite achieves a remarkable gauge factor of 675.7 (R-2 similar to 0.993) at low strain (similar to 0-4.5 %), representing a 3881.5 % improvement over pure MWCNT systems, alongside enhanced toughness (similar to 89.17 %) and strain tolerance (similar to 53.93 %). Meanwhile, the optimized composition ensures low rest-state resistance (similar to 13.1 Omega), minimal hysteresis (similar to 5.7 %), and robust durability over 5000 cycles at 10 % strain. As a result, the proposed sensor enables highly consistent, high-fidelity monitoring of various subtle-to-moderate biomotions. Integrated with a fine-tuned InceptionTime deep learning model, it achieves an F1-score of 98 % in classifying Dysphagia Diet Standardization Initiative (IDDSI)-standard swallowing activities, demonstrating its potential for AI-driven health monitoring applications.