神经网络用于苯乙胺衍生物的QSAR研究

Neural Networks in QSAR Studies:Estimation and Prediction of the Biological Activity for N,N-dimethyl-2-bromo(substituted phenyl) ethylamine Derivatives

将神经网络应用于定量构效关系（ＱＳＡＲ）中。采用改进的反向传播算法探讨了肾上腺素能阻断剂Ｎ，Ｎ－二甲基－２－溴苯乙胺取代衍生物的生物活性与取代基疏水参数（Σπ）和电子参数（Σσ）之间的关系。获得了精密的拟合及准确的预测（最大误差小于１０％），优于多无线性回归法。作为一种有效的化学计量学工具（Ｃｈｅｍｏｍｅｔｒｉｃｓ），神经网络具有良好的预测效果及较强的非线性处理功能，可望在ＱＳＡＲ及药物制剂中发挥重要作用。

Neural networks(NN) methods were applied to quantitative structure-activity relationship(QSAR) studies.The relationship between biological activity(PC) and eletronic and hydrophobic parameters(sigma and pi) of N,N-dimethyl-2-bromo-(substituted phenyl)-ethylamines was investigated by using the modified backpropagation(MBP) neural networks. The biological activity of N, N-dimethyl-2-bromo-(substituted phenyl)-ethylamine derivatives was estimated and predicted with the relative error less than 10% and with correct classification ratio being 91%. The results obtained by the developed MBPNN method seemed to be better than those by multivariate linear regression(MLR). The neural networks(MBPNN) method might therefore be regarded as an excellent and effective chemometric modelling technique for estimating and predicting biological activity on basic QSAR studies.