Abstract: The relationship between the affinity of 24 triphenylacrylonitrile derivatives acting on estrogen receptor in calf uterine tissue (lg1C) and X-hydroxy indicators (I)，molecular surface area (SA)，and the sum of net charge on B ring (QB) was discussed based on an improved back-propagation (BP) algorithm of artificial neural network (ANN). Selecting 20 compounds as the training set，the QSAR model was established with the ANN method. The residual 4 compounds as the prediction set were applied to test the predicted effect of the QSAR model. It was obtained that the correlation coefficient of QSAR model was R=0.9969 and the standard deviation was SD=0.0164. For the prediction set，R=0.9969 and SD=0.1533. The QSAR model for the same 24 compounds was also established with the multiple linear regression (MLR) method for comparison，with which R=0.9360 and SD=0.3779 were obtained. The results indicated that the fitted performance of ANN method is better than that of MLR model，which is comparatively precise and has a preferable predicted effect

摘要： 采用人工神经网络（ANN）BP算法探讨了24个三苯基丙烯睛衍生物的lg1/C（C为半致死浓度）与X位羟基指示数I、分子表面积SA和B环上原子净电荷之和QB之间的关系，以20个样本为训练集建立了定量结构-活性关系（QSAR）模型，其相关系数和标准偏差分别为R=0.9969和SD=0.0164，其余4个样本为测试集，得到R=0.9913和SD=0.1533；用多元线性回归（MLR）方法建立的QSAR模型R=0.9360，SD=0.3779。结果表明，ANN方法具有良好的预测能力，比MLR方法更精密。