论文部分内容阅读
目的:建立大黄素衍生物抗肿瘤活性的神经网络模型。方法:采用量子化学的AM1算法,计算了12个大黄素衍生物分子的结构参数,并用逐步回归分析,筛选结构参数。结果:利用筛选后的结构参数,建立大黄素衍生物抗肿瘤活性的神经网络模型。结论:抽一法交叉预报结果表明,本文建立的大黄素衍生物抗肿瘤活性的神经网络模型,预报结果可靠,具有一定的应用价值。
Objective: To establish a neural network model for the antitumor activity of emodin derivatives. Methods: Quantum chemistry AM1 algorithm was used to calculate the structural parameters of 12 emodin derivatives. Stepwise regression analysis was used to screen the structural parameters. RESULTS: Using the structural parameters after screening, a neural network model for the antitumor activity of emodin derivatives was established. Conclusion: The results of the method of cross-cutting by one method show that the neural network model of the anti-tumor activity of emodin derivatives established in this paper has reliable prediction results and has certain application value.