基于PCA-Logistic回归分析的艾叶药材等级评价方法研究

Research on the Grade Evaluation Method of Artemisiae argyi folium Based on PCA-Logistic Regression Analysis

目的通过分析艾叶主要有效成分,建立用于艾叶药材等级评价的“主成分(PCA)-Logistic回归”分析方法。方法取艾叶药材14批,通过气相色谱-质谱(GC-MS)法分析鉴定其挥发油各组分,通过液相色谱-四极杆-飞行时间质谱(Q-TOF LC/MS)解析黄酮类成分,选取挥发性及黄酮类有效成分中普遍存在且相对含量较高者为Logistic建模自变量,以各自变量含量测定值做PCA分析,运用Logistic二分类算法,将各成分进行关联分析,建立用于艾叶药材等级评价研究的“PCA-Logistic回归”分析方法。结果选取挥发性成分中桉油精、龙脑,黄酮类成分中异泽兰黄素、棕矢车菊素为Logistic建模自变量。桉油精、龙脑含量范围分别介于0.052%~0.092%,0.021%~0.031%,异泽兰黄素、棕矢车菊素含量范围分别介于0.189%~0.324%,0.068%~0.088%。二分类Logistic回归分析结果表明,14批次艾叶分为4个等级,各批次艾叶等级拟合概率均>95%,优级5批,良级4批,中级2批,差级3批。结论该实验基于“PCA-Logistic回归”分析,初步建立了不同批次艾叶药材等级评价新方法,可为艾叶药材等级评价标准的建立提供技术支撑。

Objective To establish a grade evaluation method of artemisiae argyi folium based on the analysis of the main active components by principal component analysis(PCA)method.Methods Fourteen batches of artemisiae argyi folium were collected.The components of its volatile oil were identified by gas chromatography-mass spectrometry(GC-MS),and the flavonoids were analyzed by liquid chromatography/quadrupole time-of-flight mass spectrometry(LC/Q-TOFMS).The most common components with higher relative content in the volatile and flavonoid active components were selected as the independent variables for logistic modeling.The PCA analysis was performed with the measured values of the contents of each independent variable,and a logistic binary classification algorithm was used to carry out association analysis on each component.Then a PCA-logistic regression analysis method for the grade evaluation of artemisiae argyi folium was established.Results Eucalyptol and borneol in volatile components,eupatilin and jaceosidin in flavonoids were selected as logistic modeling independent variables.The contents of eucalyptol and borneol were between 0.052%-0.092%and 0.021%-0.031%,respectively.The contents of eupatilin and jaceosidin were between 0.189%-0.324%and 0.068%-0.088%,respectively.The binary logistic regression analysis results showed that 14 batches of artemisiae argyi folium were divided into 4 grades,the grade probability of each batches of artemisiae argyi foliumi was higher 95%.Among the 14 batches,five batches were evaluated excellent,four were good,two were moderate,and three were poor.Conclusion Based on PCA-Logistic regression analysis,a new method was preliminarily constructed for the grade evaluation of different batches of artemisiae argyi folium,which could provide technical support for constructing the grade evaluation standard of artemisiae argyi folium.