Odor-Chemical Correlation-Based Quality Evaluation of Atractylodes macrocephala via Electronic Nose, HPLC, and Machine Learning

Atractylodis Macrocephalae Rhizoma (AMR) is a kind of traditional Chinese medicine, with the variety from Yuqian, Lin'an District, Hangzhou, considered the highest quality and termed Atractylodes macrocephalacy Yuzhu (AMY). This study examined the relationship between AMR's “scent indicates quality” principle and its chemical composition. Oil chamber analysis showed Lin'an samples had the highest density and largest chamber size. The electronic nose and machine learning analysis of 17 batches of AMR indicate that aromatic components, organic sulfides, inorganic sulfide, methane (methyl), and alkane hydrocarbon are important features, and SHAP analysis shows that the contribution of aromatic components, organic sulfides is 0.46. The results of HPLC and machine learning analysis showed that Atractylon was the most important for distinguishing AMR and AMY, with a contribution of 0.96. Six machine learning models were used to distinguish AMR and AMY, and the results showed that the accuracy of the models was high. The correlation analysis results showed that W2W had the highest correlation with Atractylon at 0.86.