Ionic liquids (ILs) are recognized as a possible replacement of traditional organic solvents, and ILs have been widely applied to extract various compounds. The present work aims to extract ginsenosides from Panax ginseng flower buds using aqueous ionic liquid based ultrasonic assisted extraction (IL-UAE). The extraction yields of 1-alkyl-3-methylimidazolium ionic liquids with different anions and alkyl chains were evaluated. The extraction parameters of eight ginsenosides were optimized by utilizing response surface methodology (RSM). The model demonstrated that a high yield of total ginsenosides could be obtained using IL-UAE, and the optimum extraction parameters were 0.23 M [C₄mim][BF₄], ultrasonic time of 23 min, temperature of extraction set to 30 °C, and liquid-solid ratio of 31:1. After that, an aqueous biphasic system (ABS) was used to separate ginsenosides further. The nature and concentration of salt, as well as the value of pH in ionic liquid were evaluated, and the optimal conditions (6.0 mL IL extract, 3 g NaH₂PO₄, and pH 5.0) were obtained. The preconcentration factor was 2.58, and extraction efficiency reached 64.53%. The results indicate that as a simple and efficient method, an IL-UAE-ABS can be considered as a promising method for extracting and separating the natural active compounds from medicinal herbs.

Traditional Chinese Medicine (TCM) has attracted more and more attention due to its remarkable effects on treating diseases, and Chinese herbal medicine (CHM) is an important partition of TCM, rich in natural active ingredients. Researchers are trying multiple analytical methods to dig out more valuable information about CHM and reveal the principle of TCM. Machine learning is playing an important role in the studies. Knowledge discovery of CHM using machine learning mainly includes quality control of CHM, network pharmacology in CHM, and medical prescriptions composed by CHM, aiming to understand TCM better, provide more efficiency methods in the production of CHM and find novel treatment of disease not curable nowadays. In this paper, we summarized the basic idea of frequently used classification and clustering machine learning algorithms, introduced pre-processing algorithms commonly used to simplify and accelerate machine learning procedure, presented current status of machine learning algorithms’ applications in knowledge discovery of CHM, discussed challenges and future trends of machine learning’s application in CHM. It is believed that the paper provides a valuable insight for the starters trying to apply machine learning in the study of CHM and catch up the recent status of related researches.