基于 BP 神经网络的半夏、生姜、甘草三泻心汤配伍研究

中国临床药理学与治疗学 ›› 2005, Vol. 10 ›› Issue (5): 527-531.

基于 BP 神经网络的半夏、生姜、甘草三泻心汤配伍研究

宋小莉, 牛欣, 司银楚¹

北京中医药大学生理教研室,¹解剖教研室, 北京 100029

收稿日期:2005-02-23 修回日期:2005-03-17 出版日期:2005-05-26 发布日期:2020-11-19
通讯作者: 宋小莉,女, 博士, 主要从事复方配伍研究。Tel:010-64488335 　E-mail:sxlsxl2004@hotmail.com
基金资助:
国家中西医结合基础重点学科 211 工程重点资助项目(No2001-004)

Study on proportion of pinellisa decoction for purging stomach-fire, gin-ger rind and licorice root with BP artificial neural network

SONG Xiao-li, NIU Xin, SI Yin-chu

Department of Physiology,¹Department of Anatomy, Beijing University of Chinese Medicine, Beijing 100029, China

Received:2005-02-23 Revised:2005-03-17 Online:2005-05-26 Published:2020-11-19

摘要/Abstract

摘要： 目的: 探讨不同 BP(Back Propagation, BP) 算法的人工神经网络在半夏泻心汤、生姜泻心汤、甘草泻心汤配伍中的应用, 并应用所建模型探讨三复方药味与剂量的配伍规律。方法: 应用均匀设计对药味及剂量进行分组, 测定不同组别对正常大鼠胃粘液的影响。应用MATLAB 6.5 进行编程, 选用 BP 神经网络来拟合实验数据, 比较 8-3-1、8-8-1、8-12-1 三种拓扑结构、不同 BP 算法对网络模型拟合效果的影响, 建立基于 BP 神经网络的三方对胃粘液含量影响的预测模型。结果: 拓扑结构为 8-8-1、算法为改进BP 算法的神经网络模型可以很好的拟合学习过的样本, 并对未学习过的样本有较好的预测能力, 其中采用动量法和学习速率自适应调整两种策略相结合的改良 BP 算法的网络拟合预测效果最佳。应用模型分析可以看出, 每种药物剂量变化及不同药物组合对胃粘液分泌的影响不尽相同, 如辛开组合具有促进胃粘液分泌的作用, 苦降组合、甘补组合具有抑制胃粘液分泌的作用。结论: 以半夏、甘草、生姜泻心汤为研究模板, 提出的复方类方配伍规律研究模式 :“优化拆方实验设计 -人工智能数据挖掘-复方类方知识发现”, 将为复杂复方的研究提供借鉴。

关键词: BP 神经网络, 半夏泻心汤, 生姜泻心汤, 甘草泻心汤, 复方配伍, 研究模式

Abstract: AIM: To observe the difference in the application of proportion of pinellisa decoction for purging stomach-fire, ginger rind and licorice root with different BP (back propagation, BP) method of artificial neural network.METHODS: The influence of gastric mucuqas content in normal rats was measured, and the groups of different flavor of a drug and dosage were divided accord-ing to homogeneous design.This research was pro-grammed with MATLAB6.5, conformed the experimental data with BP artificial neural network, and the research compared the different structural topology of 8-3-1、8-8-1、 8-12-1, and compared the different influence with the dif-ferent BP method to network.And then, the expected mode of gastric mucus content was formed, which based on the BP neural network. RESULTS: The artificial neural network model that had 8-10-1 structural topology and improved BP method was formed, and it was a kind of good method to uniform the studied samples and to forecast the unstudied samples, Moreover, the expected effect of variable learning rate back propagationwas best. CONCLUSION: Using the studying of pinellisa decoc-tion for purging stomach-fire, ginger rind and licorice as an example, and give birth to the new model in studying proportioning of compound prescription.This model will found its position in studying proportioning of compound prescription.

Key words: BP artificial neural network, proportion of pinellisa decoction for purging stomach-fire, shengjiang decotion, gancao decotion, compound prescription

中图分类号:

R289

宋小莉, 牛欣, 司银楚. 基于 BP 神经网络的半夏、生姜、甘草三泻心汤配伍研究[J]. 中国临床药理学与治疗学, 2005, 10(5): 527-531.

SONG Xiao-li, NIU Xin, SI Yin-chu. Study on proportion of pinellisa decoction for purging stomach-fire, gin-ger rind and licorice root with BP artificial neural network[J]. Chinese Journal of Clinical Pharmacology and Therapeutics, 2005, 10(5): 527-531.