Computer Engineering and Applications ›› 2020, Vol. 56 ›› Issue (16): 218-225.DOI: 10.3778/j.issn.1002-8331.1905-0279

Previous Articles     Next Articles

Gout Diagnosis Model Based on Neighborhood Cost Sensitive Three-Way Decision

ZHANG Meng, SUN Bingzhen, CHU Xiaoli   

  1. School of Economics and Management, Xidian University, Xi’an 710071, China
  • Online:2020-08-15 Published:2020-08-11

基于邻域代价敏感三支决策的痛风诊断模型

张萌,孙秉珍,楚晓丽   

  1. 西安电子科技大学 经济管理学院,西安 710071

Abstract:

The three-way decision-making is based on cost sensitivity. By introducing delayed decision-making, the classification can be more reasonable under the condition of incomplete information. In this paper, the optimization decision problem of decision information system with mixed attribute characteristics is considered. The neighborhood relationship is defined on the mixed attribute information system, and the decision rough set model based on neighborhood relation is constructed. On this basis, it is applied to the decision-making problem of gout clinical diagnosis, and the gout data are classified by iterative learning. Compared with SVM(Support Vector Machine), RF(Random Forest), LR(Logistic Regression) classification algorithms, the superiority of this method is proved. Finally, according to the classification results, this paper finds the internal relationship between the factors, obtains the classification rules, and explores the correlation between gout and liver function, kidney function, blood fat and blood sugar, so as to provide knowledge support and decision support for the cause research, diagnosis and treatment of gout.

Key words: cost sensitivity, three-way decision, neighborhood, big data medical treatment

摘要:

三支决策基于代价敏感,通过引入延迟决策,在信息不完备的情况下,能够使分类更加合理。考虑具有混合属性特征的决策信息系统优化决策问题,在混合属性信息系统上定义了邻域关系,构建了基于邻域关系的决策粗糙集模型。在此基础上将其应用于痛风临床诊断决策问题,运用多次迭代学习的方法对痛风数据进行分类。与SVM(Support Vector Machine)、RF(Random Forest)、LR(Logistic Regression)分类算法进行对比,证明了该方法的优越性。根据分类结果发现因素之间的内在联系,获取分类规则,探究痛风与肝功、肾功、血脂、血糖的相关性,为痛风成因研究和诊断治疗提供知识支持和决策支持。

关键词: 代价敏感, 三支决策, 邻域, 大数据医疗