- Article
Rapid Identification of Trace Pharmacodynamic Substances in Traditional Chinese Medicine via SERS and Deep Learning
- Huixuan Yang,
- Mingyuan Chen and
- Guochao Shi
- + 7 authors
In the modernization of traditional Chinese medicine (TCM), trace detection of pharmacodynamic substances faces a critical challenge: insufficient sensitivity, which significantly hinders accurate quality assessment and standardization. Conventional techniques often fail to measure trace components in complex sample matrices. Therefore, the development of a rapid, effective, sensitive, and reliable analytical method, along with a corresponding quality evaluation system, is of great importance. This study used moth wing (MW) scales as a template to fabricate an Ag/MW SERS substrate via magnetron sputtering. The optimal Ag30/MW SERS substrate (30 min sputtering) achieved an enhancement factor of 6.47 × 106 and good reproducibility (minimum RSD: 7.03%). Principal component analysis (PCA) was integrated with four deep learning algorithms (MLP, Transformer, ResNet, DNN) to detect three typical TCM pharmacodynamic substances in pure standard solutions: atractylon, cimifugin, and timosaponin A-III. The models enabled rapid identification, with the MLP model reaching 95.00% accuracy. This research provides a novel, highly accurate, and efficient detection method with potential for TCM pharmacodynamic substances, demonstrating feasibility for bioactive compound identification in model systems, and shows promising potential for future application in TCM composition analysis and quality control.
27 February 2026




![Schemes for sensor chip surfaces and ligand immobilization strategies in SPR for analysis of putative drugs. (A,B) An example of a sensor chip surface with dextran matrix for (A) amine-coupling immobilization of amine-containing molecule with available COO− grups or (B) with already immobilized strepavidine, NTA-Ni2+ and protein A for immobilization of biotinylated, Hist-tag-containg molecules and antibodies, respectively. (C) Analysis of interaction between the putative drug and lipid monolayer or liposomes immobilized on HPA or L1 sensor chips, respectively. (D) Left panel: host receptor protein (e.g., ACE2) immobilized via amine groups and analysis of surface binding by viral protein alone (e.g., SARS-CoV-2 spike protein). Right panel: analysis of binding immobilized host receptor by viral protein preincubated with the tested inhibitor. An inhibitor that binds to the viral protein at the same interface as the host receptor blocks the receptor’s binding. (E) Left panel: analysis of viral RNA polymerase (RdRp) binding to immobilized RNA molecule alone. Right panel: analysis of viral RdRp binding after the preincubation with the putative inhibitor. (based on [31]). (F,G) Sensorgrams showing the response after injections of the analyte onto the surface with immobilized ligand, followed by injection of the inhibitor (blue) or buffer (green). The increase in signal after inhibitor injection indicates binding to the analyte (blue) (F). The decrease in signal after inhibitor injection, more significant than after buffer injection, indicates removal of the analyte from the complex with the ligand (blue) (G). Not drawn in scale.](https://mdpi-res.com/cdn-cgi/image/w=281,h=192/https://mdpi-res.com/biosensors/biosensors-16-00136/article_deploy/html/images/biosensors-16-00136-g001-550.jpg)


