Related Experiment Video
Updated: Jan 17, 2026
05:32
A Detailed Protocol for Perspiration Monitoring Using a Novel, Small, Wireless Device
Published on: November 24, 2016
8.2K
Label-Free SERS Platform Assisted by Machine Learning for Multi-Target Detection and Physiological State
Banglei Zhu1, Jin Chen2, Bingwei Wang2
1Military Medical Sciences Academy, Academy of Military Sciences, Tianjin 300050, China.
Analytical Chemistry
|September 16, 2025
Summary
This study presents a novel label-free surface-enhanced Raman spectroscopy (SERS) method with machine learning (ML) for simultaneous sweat metabolite detection and physiological state classification, advancing personalized health monitoring.
Area of Science:
- Biosensing and Diagnostics
- Analytical Chemistry
- Machine Learning in Healthcare
Background:
- Sweat metabolite detection is vital for health monitoring but faces challenges with sensitivity, stability, and multitarget analysis.
- Traditional methods struggle with low concentrations, complex matrices, and limited multiplexing capabilities.
Purpose of the Study:
- To develop an innovative, label-free SERS method integrated with ML for simultaneous quantitative detection of glucose, uric acid, and lactate in sweat.
- To enable classification of physiological states using sweat metabolite profiles.
- To overcome limitations of traditional methods in sensitivity, accuracy, and reliability for multitarget detection.
Main Methods:
- Utilized a portable Raman spectrometer with nanostructure-enhanced SERS for amplified signal detection.
- Employed label-free detection to mitigate interference from complex biological matrices.
- Applied seven machine learning models (KNN, SVM, CNN, DNN, etc.) for quantitative analysis and physiological state classification.
Main Results:
- Achieved simultaneous quantitative detection of glucose, uric acid, and lactate in real sweat.
- The KNN model showed optimal performance for metabolite detection.
- The SVM model demonstrated 94.7% accuracy and 94.5% F1 score for physiological state classification.
Conclusions:
- The integrated SERS-ML approach significantly enhances sensitivity, accuracy, and reliability for multitarget sweat metabolite detection.
- This method offers a powerful tool for health assessments, disease screening, and personalized health management.
- Advances biosensing technologies for clinical applications and personalized medicine.
