基于粗糙集和遗传算法的中医方证相关性研究

STUDY ON THE RELATIVITY OF PRESCRIPTION AND SYNDROME IN TCM BASED ON ROUGH SET AND GENETIC ALGORITHM

针对中医方证相关性分析中存在的模糊性和不确定性问题,提出一种基于粗糙集和遗传算法的中医方证相关性分析模型。该模型使用遗传算法对方证决策表进行属性约简,构造的适应度函数一方面最大化决策属性对条件属性的依赖度,另一方面通过计算个体稀疏度与期望稀疏度的KL散度,对包含属性较多的个体进行惩罚,以便在最大化分类能力的同时获得最小的属性约简。使用该方法建立了心率失常和心绞痛的方证决策模型。实验结果表明,该方法得到的属性约简能够反映两种类型的冠心病在中药用药方面的差异,所提取的方证规则与目前中医治疗冠心病的用药原则基本一致,说明该方法用于中医方证相关性分析是可行和有效的。

Aiming at the ambiguity and uncertainty existing in the correlation analysis of traditional Chinese medicine prescriptions,a correlation analysis model based on rough set and genetic algorithm was proposed. The model used the genetic algorithm for the attribute verification of the opponent ＇s decision table,and the constructed fitness function maximized the dependence of the decision attribute on the condition attribute. On the other hand,by calculating the KL divergence of the individual sparsity degree and the expected sparsity degree,individuals with more attributes were punished in order to obtain the smallest attribute reduction while maximizing the classification ability. This method was used to establish a decision making model for arrhythmia and angina pectoris. The experimental results show that the attribute reduction obtained by this method can reflect the differences between the two types of coronary heart disease in traditional Chinese medicine. The extracted prescription rules are basically consistent with the current principles of Chinese medicine in the treatment of coronary heart disease,which shows that the method is feasible and effective for the correlation analysis of traditional Chinese medicine prescriptions.