基于UHPLC-Q-Orbitrap HRMS的赤芍饮片产地溯源研究OA

This study was based on ultra-high performance liquid chromatography quadrupole electro-static field orbital trap high-resolution mass spectrometry(UHPLC-Q-Orbitrap HRMS)technology,combined with multiple machine learning methods,to construct a production identification strategy for Sichuan Paeoniae Radix Rubra and other Paeoniae Radix Rubra slices from northern regions.The chemical composition information of the sample was obtained through UHPLC-Q-Orbitrap HRMS,and differential compounds were screened using multivariate statistical analysis.Based on this,ran-dom forest(RF),extreme gradient boosting(XGBoost),and adaptive boosting(AdaBoost)classifica-tion models were established.The performance of the model was evaluated based on the area under the working characteristic curve(AUC)of the subjects,and the Shapley Additive Explanations(SHAP)method was introduced to quantitatively rank and interpret the contribution of endogenous components in the discrimination process.The results showed that a total of 95 chemical components were detected,of which 40 showed significant differences between different production areas.Com-parison of model performance showed that the RF model outperforms XGBoost and AdaBoost in both accuracy and stability.SHAP analysis further revealed that linolenic acid,renoside,and pyridoxine contribute the most significantly in origin discrimination,and their content were significantly higher in Sichuan Paeoniae Radix Rubra,with the potential to serve as characteristic components of origin.The methodological framework established in this study can effectively achieve precise identification of Sichuan Paeoniae Radix Rubra decoction pieces,providing new ideas and technical support for the authentic evaluation and quality control of traditional Chinese medicine.