基于计算智能的心系基础证量化诊断方法学研究

The digitization and intelligentization of TCM syndrome differentiation is an important research project among the studies of TCM diagnosis. The existing researches mainly value the patterns of disease according to syndrome differentiation which were achieved by pattern matching of various syndromes. However, this process does not conform to the TCM idea of “viscera are detected via manifestations and syndromes are differentiated according to symptoms”, for which reason the above method is difficult to be effectively applied to clinical practice. In order to establish scientific and standardized digital models for syndrome differentiation, the following problems need resolving: 1. The standardization of syndrome differentiation should take the unified and standardized syndrome differentiation system as the theoretical basis. 2. For the problem of the acquisition and expression of TCM knowledge, mathematical model conforming to TCM thought of syndrome differentiation should be established. Therefore, this program takes the syndrome differentiation of five-zang system as the theoretical basis. Starting from the heart system, the diagnostic model of “symptoms→disease locations, features→syndromes” was established by applying multi-computational intelligence method. 1. By studying the literature, consulting experts and investigating epidemiology, basic syndromes, disease locations and features of heart system were explored. 2. Diagnostic models of fuzzy multi-attribute decision-making and genetic neural network were respectively established. After numerical verification and expert evaluation, the optimal models were selected. 3. The implicit knowledge within the models was explored and the intelligent syndrome differentiation system was designed to assist clinical syndrome differentiation. This project explores the principle and law of syndrome differentiation from multiple perspectives of theoretical principle, mathematical model and software application, which has profound significance in realizing the standardization, digitization and intelligentization of TCM syndrome differentiation.

辨证数字化、智能化是中医诊断领域的重要研究课题。现有研究多关注疾病辨证分型，通过对疾病各个证型的模式匹配来完成辨证，该过程不符合中医“以象测藏，从症辨证”思维，难以有效应用到临床。建立科学规范的数字化辨证模型需要解决：①辨证本身的规范化问题，要以统一、规范的辨证体系作为理论基础；②中医知识获取及表达问题，要建立符合中医辨证思维的数学模型。因此，本项目以五脏系统辨证体系为理论基础，从心系入手，尝试利用多计算智能方法协同建立“症状→病位、病性→证”的诊断模型：①利用文献研究、专家咨询和流行病学调查，探明心系基础证及病位、病性特征；②分别建立心系基础证模糊多属性决策和遗传神经网络诊断模型，经数值验证和专家评价，筛选出最优模型；③探索模型内部的隐式知识，并设计智能辨证系统，辅助临床辨证。本项目从理论原理、数学模型到方法应用多角度探索辨证的原理和规律，对实现辨证的规范化、数字化、智能化具有深远意义。

辨证论治是中医学的特色，其中辨证是临床立法、处方、用药的前提。辨证具有模糊性和复杂性特点，这给中医辨证数字化和智能化研究带来障碍。本项目以五脏系统辨证体系为理论基础，从心系入手，利用多计算智能方法协同建立量化诊断模型，①利用文献研究、专家咨询和流行病学调查，探明了心系基础证及病位、病性特征；②利用加权相似度、集成学习、模糊逻辑、复杂网络、图论等建立多种量化诊断模型，有效模拟了临床信息与心系基础证之间的复杂映射关系；③设计了心系基础证智能辨证系统，并在临床开展心系病证辅助诊断，应用效果良好。本项目从理论原理、数学模型到方法应用多角度探索心系基础证辨证的原理和规律，对实现辨证的规范化、数字化、智能化具有参考价值。