Machine learning-assisted wearable SERS-microfluidic system for rapid detection of sweat biomarkers and stimulants

Sweat is an ideal biofluid for noninvasive monitoring of athletes' physiological status during exercise, as its rich chemical composition directly reflects health and metabolic state. Herein, we develop a wearable device based on segmented circular microfluidic paper-based surface-enhanced Raman scattering (SCMP-SERS) for real-time sweat analysis during physical activity. The core sensor incorporates gold@carbon dots-based plasmonic paper, which exhibits exceptional SERS performance with a maximum enhancement factor of up to 3.6 × 10⁹. Combined with a flow-regulating microfluidic network, the system allows controlled sweat collection and quantitative analysis. It can identify key sweat biomarkers, such as urea, uric acid, and lactic acid, and distinguish differences between individuals. Furthermore, the device's multi-channel design facilitates continuous sweat sampling and longitudinal SERS quantitation throughout extended exercise sessions. Furthermore, by integrating machine learning, the SCMP-SERS device successfully identifies stimulants in artificial sweat with over 97.9% accuracy. This innovative approach demonstrates significant potential for real-time antidoping surveillance and personalized sports physiology monitoring.