基于有向有环图的慢性胃炎中医虚证症状选择及证候模型构建

Syndrome model establishment and feature selection of TCM deficiency symptom of chronic gastritis based on directed cyclic graph

目的:探讨有向有环图(DCG)在慢性胃炎中医虚证症状选择及证候模型构建中的可行性。方法:选取脾胃气虚证431例和脾胃虚寒证31例,运用粗糙集(RS)与互信息(MI)方法结合进行特征选择,然后运用DCG进行模型构建。结果:挑选出脾胃气虚证相关症状(体征)19个,脾胃虚寒证相关症状(体征)17个,筛选的相关症状基本与中医理论相符。基于筛选的症状,运用DCG方法构建证候模型,得到两种证候的分类准确率分别为脾胃气虚证74.1%和脾胃虚寒证96.0%,结果均优于朴素贝叶斯网络(NB)方法的72.1%和95.2%。结论:本研究较好地筛选了不同证候的相关症状(体征),应用考虑症状之间关系的DCG方法构建了虚证证候模型,取得了较好的分类效果,值得推广研究。

Objective： To discuss the feasibility of directed cyclic graph（DCG） in syndrome model establishment and feature selection of TCM deficiency symptom of chronic gastritis（CG）. Methods： Four hundred and thirty-one cases of syndrome of deficiency of spleen qi and stomach qi along with 31 cases of syndrome of deficient cold of spleen and stomach were selected for model establishment by DCG after feature selection by combined rough set and mutual information（RS-MI） method. Results： 19 relevant variables of the symptoms（or signs） of the spleen-stomach qi deficiency and 17 relevant variables of the symptoms（or signs） of the spleen-stomach yang deficiency were selected by the RS-MI method, which accorded with the TCM theory. The classification accuracies of both syndromes determined by DCG method were 74.1% and 96.0% respectively, which was superior to the Naive Bayes（NB） network model in terms of classification ability. Conclusion： Related symptoms of different syndromes were screened accurately in this study. Deficiency syndrome model established by DCG considering the connection between different syndromes could achieve better classification, which worth further researching and promoting.