Surface-enhanced Raman scattering integrating with machine learning for green tea storage time identification.

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Chemometrics

Background:

• Accurate identification of complex samples with similar compositions is challenging.
• Green tea quality is affected by storage time, altering its sensory components.
• Developing rapid and reliable analytical methods is crucial for food quality control.

Purpose of the Study:

• To develop an integrated strategy using surface-enhanced Raman scattering (SERS) and machine learning for discriminating green tea products based on storage duration.
• To investigate the changes in green tea's sensory components over time using SERS.
• To establish a robust classification model for identifying green tea with different storage histories.

Main Methods:

• Utilized surface-functionalized silver nanoparticles (NPs) as SERS substrates.
• Applied principal components analysis (PCA) for feature extraction from SERS spectra.
• Employed support vector machine (SVM) classification for discriminating green tea samples.
• Analyzed spectral data to correlate with storage time variations.

Main Results:

• SERS successfully revealed changes in green tea's sensory components influenced by storage time.
• PCA-SVM effectively extracted key spectral features for classification.
• The multiclass SVM classifier achieved high predictive accuracy (95.9%), sensitivity (96.6%), and specificity (98.8%) in identifying green tea by storage time.

Conclusions:

• The integrated SERS-based PCA-SVM platform provides a facile and reliable method for identifying complex matrices with subtle differences.
• This approach demonstrates significant potential for quality control and authentication of food products like green tea.
• The study highlights the power of combining advanced spectroscopic techniques with machine learning for challenging analytical tasks.