Dual\nDetection of Urea and Glucose in Sweat Using\na Portable Microfluidic SERS Sensor with Silver Nano-Tripods and 1D-CNN\nModel Analysis

Sweat, a noninvasive metabolic product\nof normal physiological\nresponses, offers valuable clinical insights into body conditions\nwithout causing harm. Key components in sweat, such as urea and glucose,\nare closely linked to kidney function and blood glucose levels. Portable\nsweat sensors, equipped with diverse sensing systems, can monitor\nfluctuations in urea and glucose concentrations, thus providing methods\nfor assessing kidney function and monitoring diabetes. This study\npresents a flexible, portable microfluidic surface-enhanced Raman\nscattering (SERS) sensor designed to detect the unique fingerprint\nof target biomarkers. This flexible, self-adhesive microfluidic chip,\nconstructed from modified polydimethylsiloxane, features silver nanotripods\n(AgNTs) with densely distributed “hotspots” created\nvia the oblique angle deposition technique. These AgNTs act as active\nsubstrates for SERS within the microfluidic platform, enabling direct\nskin contact to collect, transport, store, and analyze sweat. The\nchip functions as a quantitative urea sensor with a limit of detection\n(LOD) of 10<sup>–7</sup> M. For enhanced sensitivity for glucose\ndetection, the SERS substrate is modified with 4-mercaptophenylboronic\nacid, achieving a LOD of 10<sup>–7</sup> M. This satisfies\nthe measurement requirements for both urea and glucose in sweat under\nphysiological conditions. Furthermore, the one-dimensional convolutional\nneural network model significantly enhances the accuracy of biomarker\ndetection, facilitating the quantitative analysis of urea and glucose.\nThis advancement contributes to the development of a controlled, convenient,\nand dynamic biosensing system for personalized point-of-care testing\nand supports the creation of intelligent wearable and nondestructive\ndevices.