Multi-dimensional information characterization of different grades of Atractylodis macrocephalae Rhizoma based on HS-GC-MS, LC-MS, electronic nose, and electronic tongue.

Atractylodes macrocephala Rhizoma. (AMR, called Baizhu in Chinese) is used in traditional Chinese medicine (TCM) for treating gastrointestinal disorders, such as diarrhea and gastritis. With increasing demand, its cultivation areas have expanded significantly. However, the quality of AMR varies considerably by geographical origin and commercial grades, and current research remains insufficient regarding chemical profile differences between grades and rapid classification methods. In this study, we employed Headspace Gas Chromatography-Mass Spectrometry (HS-GC-MS) and Liquid Chromatography-Mass Spectrometry (LC-MS) to analyze volatile and non-volatile components in different grades of AMR. Multivariate statistical methods were applied to elucidate the compositional variation patterns across grades, identify components highly correlated with grading standards. Additionally, the feasibility of applying electronic nose (E-nose) and electronic tongue (E-tongue) technologies for rapid grade classification of AMR was explored. Furthermore, a multi-technology fusion strategy integrating data from HS-GC-MS, LC-MS, E-nose, and E-tongue was implemented to establish a comprehensive grading model the study revealed that there were four volatile differential compounds and six non-volatile differential compounds common to all grades of AMR. Spearman correlation analysis identified terpenoids as the primary volatile compounds contributing to grade-specific aromas, with esters and phenolic acids being key taste compounds. Comparative analysis showed that the multi-technology fusion model, particularly using the Random Forest algorithm, achieved superior classification accuracy (up to 98.33%) compared to models based on any single technology. This study establishes a robust multi-dimensional approach that enhances the quality evaluation research on AMR grading and provides a novel and more reliable strategy for rapid classification of different AMR grades.