Exploring efficacy differences and quality evaluation of Gentiana rigescens Franch. from different Jiuzhi methods...

The processing (paozhi) of traditional Chinese medicines profoundly influences their efficacy. Wine-roasted (Jiuzhi) can alter the medicinal properties of Gentiana rigescens Franch. (GRF), yet few studies have explored the effects of different wine products on these properties. This research obtained 11 Jiuzhi samples. Fourier Transform near infrared spectroscopy (FT-NIR) and Fourier Transform infrared spectroscopy (FT-MIR) and High Performance Liquid Chromatography (HPLC) techniques were employed to obtain relevant information, investigating how different Jiuzhi affect the chemical composition of GRF. Network pharmacology analysis was used to predict pharmacological differences among the processed samples. Additionally, molecular docking was employed to predict the binding conformations between the chemical components and the predicted protein receptors. Partial Least Squares-Discriminant Analysis (PLS-DA) and Random Forest (RF) models were applied to distinguish different Jiuzhi GRF. Partial Least Squares Regression (PLSR), RF, and Long Short-Term Memory (LSTM) algorithms were utilized to predict the levels of seven chemical components in the GRF. Results indicate that different Jiuzhi methods alter the content of various chemical components. The binding energies of all docked molecules were below -5 kcal/mol. PLS-DA emerged as the optimal model for discriminating between processed samples, achieving accuracy rates exceeding 90% in both train and test datasets. Among seven components, the LSTM-based Isovitexin prediction model showed the highest performance, with R² > 0.9 in both calibration and prediction datasets. This study establishes a workflow for the rapid screening of quality markers and the prediction of pharmacological mechanisms in TCMs using FTIR-based chemometric analysis and network pharmacology.