Abstract: This review summarizes the application of machine learning in research on the numbing substances of Zanthoxylum bungeanum. Different varieties of Z. bungeanum on the market vary in the composition and content of numbing substances. Traditional methods for the detection and analysis of the composition and content of numbing substances in Z. bungeanum have many limitations, so the introduction of machine learning algorithms has brought new possibilities to this field. Applying machine learning for predictive modeling of the quality of Z. bungeanum and modeling of its drying process, the sensory evaluation of Z. bungeanum, and establishing a germplasm bank for Z. bungeanum is of great significance for the genetic breeding of Z. bungeanum. This article systematically reviews the composition and content of numbing substances in different varieties of Z. bungeanum, and analyzes the application of machine learning algorithms in predictive quality modeling and drying modeling of Z. bungeanum and data analysis of its numbing substances. Integration with machine learning technology enables a deeper understanding of the currently established models, which will provide support for the optimization of the yield and quality of Z. bungeanum based on numbing substances.

Key words: machine learning; Zanthoxylum bungeanum; numbing substances; model