目的 基于"成分反映活性，活性指向功效"的中药质量控制研究思路，建立用于柴胡Bupleurum chinense等级评价的二分类Logistic模型，并开发出一种柴胡等级评价计算器。方法 采用二分类Logistic算法将柴胡的柴胡皂苷a、d含量、环氧化酶2（cyclooxygenase-2，COX-2）生物效价、ABTS⁺清除率、DPPH∙清除率和羟自由基清除能力等指标进行关联分析，建立用于柴胡的等级预测模型，并进行验证。采用多元回归分析方法、二分类Logistic回归模型结合柴胡成分-药效相关数据，建立等级评价方程。采用JavaScript和CSS3技术，将等级评价方程嵌入HTML进行计算器插件的开发。结果Logistic预测模型概率公式初步将15个产地来源柴胡最终分为优、良、中、差4级，各批次所对应的预测归属等级明确，拟合概率P值均大于90%。开发了一种无需将数据传给服务器端，通过动态地输入测定的数据，可直接输出柴胡等级评价结果的计算器。结论 首次将柴胡功效-成分相结合，建立评价柴胡质量等级的方法，可为柴胡质量评价提供重要依据；并开发出一种基于"药效-成分"相关联的柴胡质量等级快速评价的计算器插件。

Objective Based on the quality control research idea of "ingredients reflect activity, activity points to efficacy", to establish a binary Logistic model for the grade evaluation of Chaihu (Bupleuri Radix, BR, 柴胡) and develop a grade evaluation calculator for medicinal materials. Methods The binary Logistic algorithm was used to correlate the contents of saikosaponins a and d, cyclooxygenase-2 (COX-2) biological potency, ABTS⁺ scavenging rate, DPPH∙ scavenging rate, and hydroxyl radical scavenging ability in BR, establish a grade prediction model for BR, and verify it. Using multiple regression analysis method and binary Logistic regression model combined with the data related to the composition-pharmacological effect of BR to establish the grade evaluation equation. The JavaScript and CSS3 techniques were used to embed the rank evaluation equation into HTML for the development of the calculator plug-in. Results The probability formula of Logistic prediction model divided BR from 15 producing areas into four grades:excellent, good, medium and poor. The predicted attribution grade corresponding to each batch was clear, and the fitting probability P value was all greater than 90%. A calculator has been developed that can directly output herb grade evaluation results by dynamically inputting the measured data without having to transfer the data to a server. Conclusion It is the first time to combine the efficacy-composition of BR to establish a method for evaluating the grade of BR, which can be used as an important basis for future quality evaluation of BR, and develope a calculator plug-in for rapid evaluation of quality grade of Chinese medicinal materials based on the correlation of "pharmacodynamics and components".

R283.6

国家自然科学基金面上项目（82173970）；陕西省科技厅重点产业创新链（群）项目（2020ZDLSF05-08）；现代农业产业技术体系建设专项资金资助（CARS-21）；名贵中药资源可持续利用能力建设项目（2060302）