【作者】 洪净；

【导师】 朱文锋；

【作者基本信息】 湖南中医学院 ， 中医诊断学， 2002， 博士

【摘要】 基于从定性描述到定量分析的科学发展一般规律，现代医学科学遵循着此规律取得了长足进步和快速发展。作为生命科学重要组成部分的中医学，纳入量化分析方法与手段，也已成为中医现代化的必然趋势。其中，中医辨证的量化研究，是当前提高中医药临床评价水平的关键科学问题。为此，引入最新发展起来的复杂事件数学分析模型贝叶斯网络，遵循中医学基本原理，在保持中医整体观特色前提下，就中医辨证量化方法进行了系统的研究。 首先，就中医辨证量化方法学的研究现状进行了系统分析。从辨证思维方法开始，总结了辨证的整体性、发散性、直觉性、形象性思维方法在中医辨证过程中的应用情况，继而分析了迄今为止不同作者在中医辨证量化研究中所用方法的优缺点及其效果。作为辨证量化的数理基础，模糊性判断、半定量方法，多因素分析、人工智能技术均已得到了大量的应用。但是，由于各自的缺陷，无一能理想地运用于中医辨证的量化研究。随着人工智能技术的逐步完善和推广，其在本研究领域的应用已成为新的发展趋势。 其次，对中医辨证思维规律与方法进行了探讨，从分析几种主要辨证方法之间的关系入手，根据导师朱文锋教授所创辨证统一体系理论，确定了辨证思维的关键环节——辨证要素，即病位与病性，讨论了辨证要素对于建立辨证统一体系的作用及其在辨证定量研究中的应用前景。因此；在症状与证候名称规范的基础上，探求症状／体征对辨证要素的贡献度，探讨症状一要素一证侯辨证统一体系的完善与实施方案，尤其是进一步将人工智能技术与之相结合，以应用于辨证量化研究。 基于以上分析和人工智能数理模型的新发展，考虑到贝叶斯网络原理与人脑思维模式，特别是辨证思维过程的良好拟合性，较深入探讨了该模型应用于中医辨证量化研究的可行性。从网络结构、概率分析到网络的自学习与经验积累，学习算法及流程图的构建，整个过程与中医辨证思维与推理具有良好的吻合性。为4b，设计构造了基于贝叶斯网络的中医量化辨证系统。 为考证该系统的实用效果，本文以一组806例肺系疾病住院病例资料为样本，分别考察不同样本数贝叶斯网络自学习和预测效果。结果表明，当样本数达到叩0时，量化辨证的预测就已达到较佳效果，与人脑辨证思维过程拟合度较好。同时，以此研究样本为基础，分析了过程中主要相关模块的工作原理与方法。本研究结果表明，不仅中医辨证量化研究可以收到很好的效果，中医学体系的量化发展也是可能的。更多还原

【Abstract】 It is a natural tendency for Traditional Chinese medicine (TCM), as an very important part of traditional medicine, to be modernized and to become a kind of science holding the properties being able to do quantitative description and analysis by the way to take in the methods and techniques of quantitative analysis. This is determined by the law, which is true for the developmental process of science from qualitative description to quantitative analysis. Medical science has quickly progressed in this direction. Meanwhile, this progress is also involved in the key step to improve the level of clinical research on TCM at present. To promote this advancing process of quantitative analysis for TCM developing, it is an important step for syndrome differentiation of TCM to be put forward with the abilities to make diagnosis quantitatively. Therefore, this research was aimed at the methodological studies systematically for syndrome differentiation. This is done through inducing in the mathematical model of Bayesian network system firstly, which was developed recently to analyze complicated events, and on the basis to keep the important property of the concept of holism undisturbed, which is one of the essential principles in the theoretic system of TCM.This work was composed of following studies.Firstly, a systematic analysis was made on the methodologies used in the past researches on syndrome differentiation of TCM with quantitative methods. Begun at the modes of thinking for syndrome differentiation, a summary was made to review the application of overall, diverging and imaging thinking modes in the process of syndrome differentiating. Following this, advantages, disadvantages and practicaleffectiveness were analyzed for quantitative methods used by various researchers in the study of quantitative syndrome differentiation. As the mathematic basis of quantitative syndrome differentiation, the principles of fuzzy determination, semi-quantitative method, single-factor analysis and multiple-factor analysis have all been applied in this field of study largely. However, no one could be used ideally in the quantitative study of syndrome differentiation, because of their own intrinsic faults. As the developing of principles of artificial intelligence and popularizing of bionic intelligence computer technique, the application of artificial (network) intelligent techniques in the field of syndrome differentiation research should be the new tendency of development.Secondly, an investigation was carried out on the laws and methods of thinking in syndrome differentiation of TCM. Started at the analysis of correlations among several main methodologies used for syndrome differentiation, syndrome differentiation union system (SDUS) was more carefully discussed based on the principles put forward by Prof. Zhu Wenfong. Here, the key elements of syndrome differentiation were taken as the key link during the thinking process of syndrome differentiation, i.e. lesion location and lesion nature. The effects, position, and its applying in future in the study of quantitative syndrome differentiation were the main subjects absorbing our attention on SDUS. So, the emphasis of this work was put on the further improvement and implementing program for such a SDUS of symptoms-key elements-syndrome, based on calculation of contribution degree for symptoms/signs to the key elements of syndrome differentiation. Particularly, this paper was focused on the combination between quantitative techniques of syndrome differentiation study and SDUS to make it developing further and for it to be used in the research of quantitative syndrome differentiation in TCM.Thirdly, an investigation was made on the applicability of Bayesian network model used in the quantitative study of TCM syndrome differentiation at the first, based on the analysis as stated above and the review on the newest development in mathematic model of artificial intelligence. The basic consideration for this was on the simulation between Bayesian network model and the th更多还原