基于遗传规划的疾病鉴别诊断研究  

Research of disease differential diagnosis based on Genetic Programming

作　　者：马桂峰[1] 张惠卿[2] 周梅[1] 代学之[2] 孙永红[3] 盛红旗[4] 陈景武[1] 孙肖伟[3] 

机构地区：[1]潍坊医学院,山东省潍坊市261053 [2]潍坊市人民医院 [3]潍坊医学院附属医院 [4]潍坊市卫生局

出　　处：《中国医院统计》2010年第4期289-292,共4页Chinese Journal of Hospital Statistics

基　　金：山东省自然科学基金项目（ZR2009CM069）潍坊市科技局2007年科技立项课题（200702044）

摘　　要：目的对难以诊断为Graves病或Hashimotos病的患者之各项临床检查指标进行定量分析，构建函数模型，以对2种疾病做出正确诊断。方法收集、整理2所三级甲等医院患者基础数据，引入遗传规划（Genetic Program—ming）法，借助MATLAB的GPLAB工具箱建立函数模型，对这2种疾病进行准确的鉴别诊断。结果建立了疾病诊断的非线性模型，在临床实践中应用效果良好，能够有效的降低误诊率。结论遗传规划法建立的非线形函数模型可以用作Graves病或Hashimotos病的鉴别诊断，有一定的推广价值和临床实践意义。Objective To establish function model through quantitative analysis of the patients＇ clinical parameters to make correct diagnosis of Graves or Hashimotos and avoid misdiagnosis. Methods Collecting and sorting the basic data of the patients from 2 top first-class hospitals. Then introduce genetic programming method to establish function model by use of MAT- LAB＇s GPLAB toolbox and introdusing the model to make accurate diagnosis for these 2 kinds of disease. Results Non-linear diagnosis model was established for the patients with disease of Graves or Hashimotos. It has been applied in the clinical practice and achieved good effect. Conclusion The Non-linear diagnosis model can tell us the disease is Graves or Hashimotos. So it has certain popularization value and clinical practice significance.

关 键 词：GRAVES病 Hashimotos病 遗传规划 误诊率 函数模型 

分 类 号：R195.1[医药卫生—卫生统计学]