目的 构建银黄颗粒中药材-中间体-成方制剂抑菌谱-效相关质量评价系统。方法 首先采用最小二乘支持向量机（LS-SVM）方法，建立银黄颗粒体外抑菌谱-效相关质量评价系统；再采用有监督的偏最小二乘判别分析（PLS-DA）方法，对黄芩药材、金银花药材、黄芩提取物、金银花提取物、银黄颗粒谱-效相关抑菌药效评价结果进行评判。结果 建立的银黄颗粒体外抑菌谱-效相关质量评价数学模型，预测结果平均相对误差在5%以下；黄芩药材、黄芩提取物、金银花药材、金银花提取物的抑菌率分别大于43%、5.5%、11%、37%，可以保证银黄颗粒87%的样品抑菌率大于11%（优质样品）。结论 建立的银黄颗粒中药材-中间体-成方制剂抑菌谱-效相关质量评价系统，能够实现中成药生产投料关键环节的质量控制，评价结果更科学、全面、准确。

Objective To establish the quality evaluation system based on spectrum-antibacterial effect correlation of Yinhuang Granules in order to detect its raw medicinal materials, extracts, and preparation simultaneously. Methods Firstly, the spectrum-antibacterial effect correlation quality evaluation system of Yinhuang Granules was established by using Least Squares Support Vector Machine (LS-SVM). Then, by using supervised partial least square-discriminant analysis models (PLS-DA), the antibacterial effect evaluation was judged based on the spectrum-antibacterial effect data of Scutellariae Radix, Lonicerae Japonicae Flos, Scutellaria extract, Lonicerae japonica extract, and Yinhuang Granules. Results The mathematical model of Yinhuang Granules based on spectrum-antibacterial effect correlation was established; The average relative error of the prediction results was less than 5%, and the antibacterial rate of Scutellariae Radix, Scutellaria Radix extract, Lonicerae Japonicae Flos, Lonicerae japonica extract was greater than 43%, 5.5%, 11%, 37% calculated by their mathematical model, which can ensure the antibacterial effect was greater than 11% (correct rate was 87%). Conclusion The quality evaluation system can realize the quality control of the key link of the production of traditional Chinese medicine, the evaluation results are more scientific, comprehensive, and accurate.

R285.5

国家重点研发计划项目（2017YFC1701501）；国家重点研发计划项目（2017YFC1702700）；山东省重点研发计划（2016GSF202005）；山东省重点研发计划（2017CXGC1306）