基于决策树的代偿期和失代偿期肝硬化自动诊断的方法  

Diagnostic Method Based on Decision Tree for Compensated and Decompensated Liver Cirrhosis

作　　者：王彦[1] 马利庄[1] 褚娜[1] 周忞[2] 

机构地区：[1]上海交通大学计算科学与工程系,上海200240 [2]上海中医药大学中医信息科学与技术中心,上海201203

出　　处：《现代生物医学进展》2008年第1期126-128,140,共4页Progress in Modern Biomedicine

基　　金：国家973项目(2006CB504801)资助;世博专题中西医信息融合的智能化综合诊断系统的资助;国家自然科学基金委创新研究群体基金项目(60521002)资助

摘　　要：决策树方法因结构简单、便于理解和具有较高的分类精度而在数据挖掘中被广泛采用。本文利用改进的决策树算法C4.5从201例肝硬化病例中自动地提取相应的肝硬化状态识别规则,得到决策树分类模型并归纳出代偿性肝硬化和失代偿性肝硬化的诊断规则,识别正确率为84.6%。实验结果表明决策树能较好的自动从肝硬化病例中归纳出代偿性和失代偿性肝硬化的诊断规则。Decision-tree calculation has been widely used in the data mining because of its simple structure and easy understanding as well as higher classification precision. In this article, by means of improved decision-tree calculation C4.5, corresponding recognition principles of hepatic cirrhosis were automatically extracted from 201 cases of hepatic cirrhosis, classified models of decision-tree was obtained, diagnostic principles for compensated and decompensated cirrhosis were induced. The accuracy rate of identification was 84.6%. The experimental results showed that diagnostic principles for compensated and decompensated cirrhosis could be automatically drawn from patients with hepatic cirrhosis by decision-tree calculation.

关 键 词：决策树 肝硬化 数据挖掘 代偿期 失代偿期 

分 类 号：R575.2[医药卫生—消化系统]