基于基因表达式编程挖掘中医方证关系

Mining formula-syndrome relationship in traditional Chinese medicine with gene expression programming

为解决中医方-证关系的函数挖掘的特殊困难,提出了基于基因表达式编程的新方法。针对中医学方证数据的特点提出结构简单、高效,占用空间少的主次分离指针编码。提出基于罚函数的单方适应度值计算和基于相似度的中医学症状组平均相似度函数,实现了基于基因表达式的方证关系函数发现算法MRAGEP,获得了一个方证函数,不但有很好的精度,而且符合中医主次证原理,可解释性好。实验证明了算法的正确性和有效性,正确率达到了62.5%

To solve the problem in mining formula-syndrome relationship of traditional Chinese medicine, a new approach based on gene expression programming was proposed. An effective and space-saving coding method was presented. The concepts of major homology（HMA） and minor homology（HMI） for formula and syndrome were defined, and a special fitness function based on penalty function method was designed, then the relationship mining algorithm based on improved gene expression programming（MRAGEP） in accordance with the formula-syndrome relationship of traditional Chinese medicine was implemented. It can get an amazing function of good precision. The experiments prove the effectiveness of MRAGEP, and its accuracy is 62.5%.