头孢菌素类抗生素的定量构动关系  

Quantitative structure-pharmacokinetics relationships of cephalosporin antibiotics

作　　者：周鲁[1] 郝丽芬[1] 

机构地区：[1]四川大学化学工程学院制药工程系,四川成都610065

出　　处：《华西药学杂志》2006年第6期534-536,共3页West China Journal of Pharmaceutical Sciences

摘　　要：目的研究头孢菌素类抗生素的定量构动关系(QSPR)。方法应用半经验量子化学方法和误差反向传播人工神经网络方法。结果利用量子化学AM1算法计算了26种头孢菌素的18个结构参数,筛选与头孢菌素类抗生素药动学指标相关性较大的6个结构参数作为人工神经网络的输入参数,以头孢菌素类抗生素的5个药动学指标(t1/2α、t1/2β、AUC、C l和Vap)作为人工神经网络的输出参数,利用误差反向传播的神经网络构建头孢菌素类抗生素的QSPR模型。从26个头孢菌素类抗生素样本中,随机抽取了23个样本作为训练集,剩余的作为检验集来验证QSPR模型的预测效果。结论头孢菌素类抗生素的ELUMO、Diop、EH、LogP、M和HDO是影响其药动学指标的主要因素,所建立的QSPR模型能够有效地描述并预测头孢菌素类抗生素的药动学指标。OBJECTIVE To study on the quantitative structure -pharmacokinetics relationships（QSPR） of cephalosporin antibiotics. METHODS The method of semi - empirical quantum chemistry and back - propagation network were used. RESULTS 18 molecular descriptors of the 26 cephalosporin antibiotics have been calculated. And 6 molecular descriptors among them, have been selected for the inputs of the back - propagation network, while the outputs of the network were the pharmacokinetics （PK） parameters （ t1/2α, t1/2β, AUC, Cl and Vap ）. 23 cephalosporin antibiotics were selected stochastically as the training set and the residual 3 cephalosporin antibiotics as the prediction set for testing the predicted effect of the QSPR model. CONCLUSION ELVMO, Diop, EH, LogP, M and HDO are the main factors on affecting the PK parameters of cephalosporin antibiotics. The QSPR model can be effectively predict the PK parameters of cephalosporin antibiotics.

关 键 词：头孢菌素类抗生素 量子化学 神经网络 定量构动关系 

分 类 号：R96[医药卫生—药理学]