基于二分类Logistic回归分析的桂枝等级预测研究

目的结合质量控制成分和生物活性对桂枝药材质量等级的影响,建立一种基于"成分-功效"研究思路的二分类模型,为桂枝的质量分级提供依据。方法采用超高效液相色谱法(UPLC),建立质控成分的含量测定方法,以DPPH和羟自由基清除实验来反映桂枝药材体外抗氧化活性,采用Logistic算法,将质控指标和抗氧化指标进行关联分析,最后建立用于桂枝分级的二元Logistic回归模型。结果建立了20批桂枝样品的UPLC指纹图谱,并对其抗氧化活性进行测定。采用主成分分析筛选出了4个质量控制成分香豆素、桂皮醇、肉桂酸、桂皮醛,并对其进行方法学验证。根据回归方程,初步将20批桂枝药材分为优、良、中、差4级。结论基于二元Logistic回归模型来描述桂枝饮片等级与影响因素之间的映射关系是可行的,可以更好地表达投料饮片等级分类标准,为中药桂枝的质量评价标准制定提供了新的思路。

Prediction of Cinnamomum cassia grade based on binary Logistic regression analysis

Objective In this study, a two-classification model based on the idea of "ingredient-efficacy" was established for the quality classification of Cinnamomum cassia with considerations to quality control components and biological activities. Methods A method to determine quality control components was proposed by UPLC. The in vitro anti-oxidant activity of C. cassia was reflected by DPPH and hydroxyl radical scavenging experiment. The quality control index and anti-oxidant index were correlated by a Logistic algorithm. Finally, a binary logistic regression model for classification of C. cassia was established. Results UPLC fingerprints of 20 samples of C. cassia were established, and their anti-oxidant activities were determined. Four quality control components (coumarin, cinnamyl alcohol, cinnamic acid, and cinnamaldehyde) were screened out by principal component analysis, and their methodological validation was carried out. According to the regression equation, 20 batches of C. cassia were divided into four grades:excellent, good, medium, and poor. Conclusion The binary logistic regression model can describe the mapping relationship between the grade of C. cassia. It can better express the classification standard for the prepared C. cassia. This study provides a new idea for quality evaluation of C. cassia.