摘要
中医证候和症状描述错综复杂,如何较好地对病患所属证候进行鉴别诊断,一直是临床医疗工作者的首要目标,把数据挖掘技术的朴素贝叶斯分类方法应用到中医证候的诊断识别中,是一个较好的尝试.为了提高分类识别的效率,在分类特征的选择上,使用了遗传算法对原有的特征进行了优化.在使用朴素贝叶斯分类方法对中医证候进行分类识别并用遗传算法改进时,经历了以下过程:首先合理抽象鉴别诊断过程并建立数学模型;其次,提出了使用数据挖掘技术中的朴素贝叶斯分类方法对模型求解;第三,考虑到特征数量较大,运用了遗传算法进行特征优化;最后,使用医学上常用的ROC曲线评价方法对改进前后的分类识别的效率进行分析比较.
The syndrome description of the Traditional Chinese Medicine (TCM) is complicated,how to carry on the distinction diagnosis well for sick syndrome, is always the clinical medical worker's priority target. The naive bayesian classification method of the data mining technology applying to the diagnosis recognition of the syndrome in TCM is a good attempt. In order to improve the efficiency of the classification, in the classified characteristic choice, the genetic algorithm is used to carry on the optimization for the original characteristic. When the naive bayesian classification method carries on the classified recognition to the syndrome in TCM and with the genetic algorithm improvement, the following processes have been experienced. First, abstract reasonably distinction diagnosis process and establish the mathematical model;Next,use the naive bayesian classification method of the data mining technology to the model solution; Third, consider the large number of characters, the genetic algorithm is utilized to carry on the characteristic optimization; Finally,ROC curve assessment method is used commonly in medicine to analyze and compare the efficiency of classified recognition around improvement.
出处
《内蒙古大学学报(自然科学版)》
CAS
CSCD
北大核心
2007年第5期568-571,共4页
Journal of Inner Mongolia University:Natural Science Edition
基金
内蒙古自然基金资助项目(项目编号:200508010814)
关键词
朴素贝叶斯分类
遗传算法
ROC曲线
中医证候
Naive bayesian classification
genetic algorithm
ROC curve
the syndrome in TCM