烟台大学学报（自然科学与工程版）2025，Vol.38Issue(2)：217-223,7.DOI:10.13951/j.cnki.37-1213/n.231106
成分表征结合机器学习的木瓜药材鉴定及化学标志物筛选
Identification of Chaenomelis Fructus and Screening of Chemical Markers by Component Characterization Combined with Machine Learning
摘要
Abstract
A method for identification and differentiation of two common Chaenomeles species in China and South Korea—Chaenomeles speciosa and Chaenomeles sinensis—was established using component characterization com-bined with machine learning models.Liquid chromatography-mass spectrometry technology was employed to deter-mine the content of 14 major components in the metabolic spectra of both Chaenomeles species,and a random forest model was used to achieve their accurate classification.Feature selection and comparison were conducted through mean decrease Gini and area under the curve,identifying six chemical markers for distinguishing the two types of Chaenomeles medicinal materials:chlorogenic acid,caffeic acid,rutin,quercetin,betulinic acid,and ursolic acid.Based on this,the extreme gradient boosting algorithm was used to derive a simplified discrimination criterion for the Chaenomeles:the content of chlorogenic acid(below 50.28 mg/kg indicates C.sinensis,otherwise it is C.speciosa),which showed good predictive performance on the test dataset.This provides a simple and quantifiable novel strategy for the identication of easily confused traditional Chinese medicine.关键词
木瓜药材/成分表征/机器学习/鉴定/化学标志物Key words
Chaenomelis Fructus/component characterization/machine learning/identification/chemical marker分类
医药卫生引用本文复制引用
李震,丁晨,李玉,赵妍..成分表征结合机器学习的木瓜药材鉴定及化学标志物筛选[J].烟台大学学报（自然科学与工程版）,2025,38(2):217-223,7.基金项目
山东省自然科学基金资助项目(ZR2023QH054) （ZR2023QH054）
烟台大学研究生科研创新基金资助项目(GGIFYTU2335). （GGIFYTU2335）
