Identification of Different Steaming Degrees of Panax quinquefolius L. Based on GC-IMS Combined With Machine Learning.

Rationale: Panax quinquefolius L. (PQ), a commonly used traditional Chinese medicine and a food, is usually processed into various products, including white PQ, red PQ (two- or three-time steamed PQ), and black PQ (nine-time steamed PQ). Previous studies demonstrated that volatile components (VOCs) were the important active substances of PQ, which had antibacterial, antiviral, and anti-leukemia activities. However, most research had focused on ginsenosides, and few studies on the volatile components (VOCs) of PQ.

Methods: This study used gas chromatography-ion mobility spectrometry to analyze the variation of VOCs in PQ during steaming process. Further, machine learning algorithms were used to quickly identify the steaming degrees of PQ samples.

Results: A total of 58 VOCs were identified, and 20 featured components with significant changes in the content were screened, including 2-methylundecanal, n-propanol, and n-octanol. Based on these 20 featured components, six machine learning algorithms were used to predict PQ samples with different steaming degrees. Among them, naive Bayes (NB) and linear discriminant analysis (LDA) exhibited good predictive performance, demonstrating significant potential application. This study provided a reference for understanding the variation of VOCs in PQ during steaming and offered a simple, rapid, and low-cost method for distinguishing the steaming degrees of PQ samples.