【作者】 阴小雄；

【导师】 唐常杰；

【作者基本信息】 四川大学 ， 计算机应用技术， 2005， 硕士

【摘要】 中国医药学具有数千年的悠久历史。它是中华民族长期以来与疾病作斗争的智慧结晶,是我们优秀民族文化遗产中的一颗璀灿明珠。千百年来,中国医药学为中华民族的繁衍昌盛和促进世界医学的发展作出了卓越的贡献。早在秦汉时代,就有了最早的本草著作《神衣本草经》。中草药的疗效不但经受住了长期医疗实践的检验,而且也已被现代科学研究所证实。大量事实证明,我国古代劳动人民通过长期实践所积累起来的医药遗产是极其丰富、极为宝贵的。我们应当珍视这个祖国医药学的伟大宝库,努力发掘。几千年的历史积累使得我国有极其浩瀚的药物和复方方剂资源。揭示方剂中单药间、方剂间的关系是当今的热点与难点。数据挖掘技术的发展和计算机性能的提高为揭示它们千丝万缕的内在联系提供了有力的技术支持。四川大学和成都中医药大学联合课题组对这一课题进行了研究,本文承担了项目中古中医复方方剂中单药、药对剂量分析工作及古中医复方方剂单药组成的频率对其配伍关系影响的预研、基于多维矢量技术的古中医复方方剂性能的可视化数据探索和聚类的预研,主要工作如下: 1 从古中医复方方剂单药组成的频率来研究单药组成对其配伍关系的影响。设计了一种古中医复方方剂的组成单药间依赖性判定判别方法(IE),在考虑支持度-置信度框架的同时,实现从关联分析到相关分析的转移。其主要特点是:(a)可以准确的表征出古中医复方方剂库DB 中方剂的组成多味单药间虽然不够频繁但却具有强依赖性依赖模式。(b)可判别出脾胃类方剂库DB 中复方方剂的组成单药的增减对该复方方剂单药间的依赖模式的影响。(c)该判别式符合复方更多还原

【Abstract】 Chinese Traditional Medicine (TCM) has a long history of several thousands years and is a result which Chinese people have struggled with the diseases. For many years, Chinese Traditional Medicine greatly has boosted the development of the world medicine and the prosperity of the Chinese Nation. A great deal of facts prove that Chinese Traditional Medicine Prescriptions has remarkable functions to treat all kinds of diseases and has been made sure by modern science research. We should be proud of our Chinese Traditional Medicine resources. This article piles up the immensity traditional Chinese Medicine Prescriptions in China for a long time. It is a hotspot and a difficulty problem to open out the relations between medicines and Chinese Traditional Medicine Prescriptions in the point view of modern science .The development of Data Mining techniques and the enhances of the computer performance offer us a powerful support to open out the inherence relations from a mass of Chinese Traditional Medicine Prescriptions. Based on the analyzing traditional research works, this thesis researches the composed rules of Chinese Traditional Medicine Prescriptions and the association of the medicines. The main contributions of this thesis include: 1 Investigates the effect of the medicine makeup frequency on the basis of medicine paring correlations of Traditional Chinese Medicine prescriptions and design a Independence Estimate method(IE). It not only thinks over the frame of support-count and confidence but also thinks over the transfer from associate analysis to correlation analysis. The main features includes: (a) It exactly captures the pattern which is not frequent enough but has a strong correlation relations in prescriptions DB. (b) It distinguishes the affect of the independence pattern while adding or decreasing the machine of Traditional Chinese Medicine Prescriptions in prescriptions DB. (c) Its discriminant satisfies the physics characteristic of Traditional Chinese Medicine Prescriptions. 2 Researches the match characteristics of the medicine in prescriptions DB under dose-change of TCM Prescriptions. By difference-Function measures the global trend of the medicine dose in DB, designs a Modifiable Difference-Function method(MDF) to search the accustomed value of the medicine in DB. The value is treated as a basic reference for missing dose medicine. In addition, it researches method for the outlier in DB. Adopts the covariance to measure the rule of the dose matching in prescriptions. 3 Starting from microcosmic view, it explores visual and digital TCM Prescriptions, designs a Multi-Vector Technique method(M-VT) to figure the performance of TCM Prescriptions. The kernel problem is to figure the performance of TCM Prescriptions. This paper adopts the dose value as map domain, gives performance of prescriptions, summarizes prescriptions performance, insures the reasonableness of the digital prescriptions, makes solid foundation for performance comparison, clustering, and classification of the prescriptions,更多还原