Machine learning-driven Ag/SiO2/Cu/rice leaf SERS platform for intelligent identification of pharmacodynamic substances

The combination of surface-enhanced Raman scattering (SERS) and machine learning algorithm provides an effective means for the identification of pharmacodynamic substances. This paper reported a biomimetic synthesis route of Ag-30/SiO2-4/Cu-20/rice leaf (Ag-30/SiO2-4/Cu-20/RL) SERS platform with multiple synergistic electromagnetic enhanced performance. The localized surface plasmon resonance (LSPR) effect was strengthened as the SiO2 nanolayer generated between Ag and Cu. This SERS platform demonstrated high sensitivity, with a low limit of detection (LOD) of 1 × 10−10 M for 4-Aminophenylthiophenol (4-ATP) and an enhancement factor (EF) of 3.86 × 106. More importantly, the principal component analysis (PCA) was adopted to analyze the SERS data of three different Traditional Chinese medicine (TCM) pharmacodynamic substances (Orientin, Atractylenolide III and Prim-o-glucosylcimifugin). The K-nearest neighbor (KNN) and Naïve Bayes (NB) achieved classification accuracy of 0.9474 and 0.9649, respectively. The platform provides guidance for the accurate identification of TCM pharmacodynamic substances.