药物的定量结构色谱保留关系研究  

Study on the Quantitative Structure Chromatographic Retention Relationship of Drugs

作　　者：何琴[1] 黄保军[1] 赵伟超 徐可宣 HE Qin;HUANG Baojun;ZHAO Weichao;XU Kexuan(School of Chemical and Materials Engineering,Xuchang University,Xuchang 461000,China)

机构地区：[1]许昌学院化工与材料学院,河南许昌461000

出　　处：《许昌学院学报》2024年第5期55-58,共4页Journal of Xuchang University

基　　金：河南省科技攻关项目(242102240089);河南省高等教育学会高等教育研究项目(2021SXHLX089)。

摘　　要：采用人工神经网络(ANN)建立了110种药物的结构与色谱保留之间的定量关系(QSRR)模型,并将其与偏最小二乘回归(PLSR)模型进行比对.110种药物的ANN模型自相容能力的复相关系数(R^(2))为0.9191,泛化能力的复相关系数(R^(2))为0.9365;而PLSR模型的复相关系数(R^(2))为0.8275和0.8514.结果表明,ANN模型的自相容能力、泛化能力优于PLSR模型.采用集内集和集外集对模型进行检验时,ANN模型的复相关系数(R^(2))为0.9105,也优于PLSR模型,且ANN预测精密度和准确度更高.The quantitative relationship(QSRR) model between the structure and the chromatographic retention of 110 drugs was established by the artificial neural network(ANN) and compared with the partial least squares regression(PLSR) model.The multiple correlation coefficient(R~2) of the ANN model self-compatibility of 110 drugs was 0.9191.The multiple correlation coefficient(R~2) of generalization ability was 0.9365.The multiple correlation coefficient(R~2) of PLSR model was 0.8275,and the multiple correlation coefficient(R~2) of generalization ability was 0.8514.The results showed that ANN model has better self-compatibility and generalization ability than PLSR model.In addition,when using the in-set and the out-of-set to test the model,the complex correlation coefficient(R~2) of ANN model is 0.9105,which is better than PLSR model,and the precision and accuracy of ANN prediction are higher.

关 键 词：定量结构色谱保留相关 人工神经网络 药物 保留因子 

分 类 号：O657.7[理学—分析化学]