Related Experiment Video
Updated: Oct 7, 2025
09:09
A Label-free Technique for the Spatio-temporal Imaging of Single Cell Secretions
Published on: November 23, 2015
8.7K
Machine Learning Methods of Regression for Plasmonic Nanoantenna Glucose Sensing
Emilio Corcione1, Diana Pfezer2, Mario Hentschel2
1Research Center SCoPE, Institute for System Dynamics, University of Stuttgart, 70563 Stuttgart, Germany.
Sensors (Basel, Switzerland)
|January 11, 2022
Summary
Gaussian process regression significantly improves glucose concentration sensing from optical spectra. This machine learning approach enhances accuracy by over 60% for biosensing applications, outperforming traditional methods.
Area of Science:
- Analytical Chemistry
- Spectroscopy
- Biosensing
Background:
- Glucose measurement is crucial for clinical applications and biosensing research.
- Surface-enhanced infrared absorption (SEIRA) spectroscopy offers sensitive and specific glucose detection.
- Accurate glucose quantification in complex solutions remains a challenge.
Purpose of the Study:
- To identify the optimal method for estimating glucose concentration in aqueous solutions containing fructose using SEIRA.
- To compare advanced machine learning regression algorithms against traditional linear regression for this inverse sensing problem.
Main Methods:
- Collected reflectance spectra from aqueous glucose and fructose solutions.
- Applied a pre-processing routine to sensor data for pattern extraction.
- Evaluated multiple machine learning regression models, including Gaussian process regression.
- Compared model performance against established linear regression techniques.
Main Results:
- Gaussian process regression demonstrated superior accuracy and reliability in predicting glucose concentrations.
- The proposed machine learning approach achieved over a 60% improvement compared to previous methods.
- Identified key patterns in SEIRA spectra for enhanced information extraction.
Conclusions:
- Machine learning regression models, particularly Gaussian process regression, are highly effective for SEIRA sensor calibration.
- Advanced algorithms offer significant improvements for glucose sensing in complex mixtures.
- Findings provide valuable insights into the capabilities and limitations of SEIRA for glucose quantification.
