Integrated Multimodal Fusion Network and Multispectral Sensing Photoelectrochemical Platform for Sweat Glucose Monitoring.

Sweat glucose detection offers the advantages of being painless, noninvasive, and suitable for continuous monitoring, showing broad application prospects in diabetes management. However, a single sweat glucose detection method cannot meet the requirements of multiscenario applications. To address this challenge, we developed a multimodal sweat glucose sensing platform integrating a type-II heterojunction with rare-earth nanoparticles (RENPs), which achieves an ultralow limit of detection (LOD) and a wide linear range while enabling robust adaptation to diverse scenarios. First, a photoelectrochemical sensing strategy was developed based on a Co(OH)2/Co3O4@Ag (CCA) type-II heterojunction, in which the surface plasmon resonance (SPR) effect of Ag nanoparticles lowered the LOD from 1.85 μM to 0.312 μM while maintaining a broad linear detection range of 0.001-13.776 mM. Second, a portable fluorescence quenching-recovery system was constructed based on the inner filter effect between RENPs and CCA, overcoming the corrosion and safety limitations inherent to alkaline electrolytes. To resolve spectral overlap issues at low concentrations, a dual-branch deep learning model was further designed to jointly process dual-modal fluorescence and absorption spectra, achieving 99.25% accuracy within the range of 10 μM-2 mM. Collectively, the integrated "RENPs-CCA-AI" strategy systematically addresses key challenges in sensitivity, detection range, wearable compatibility, and analytical accuracy, offering a systematic pathway toward noninvasive, continuous, and precise health monitoring.