满足多重分类误差的Gensini参数识别分类算法  

Gensini parameter identification classification algorithm for meeting the requirement of multiple classification errors

作　　者：方骁然 应忍冬[1] 梁伟[2] 赵雅洁[2] 

机构地区：[1]上海交通大学电子工程系,上海200240 [2]上海交通大学医学院附属瑞金医院老年科,上海200240

出　　处：《信息技术》2014年第12期1-3,8,共4页Information Technology

基　　金：上海交通大学"医工(理)交叉研究基金"项目(YG2011MS39)

摘　　要：文中研究基于常规体检数据对体检者的Gensini参数范围进行分类的方法,区分体检者分别在Gensini参数3个区间中的一个。由于判定分类性能需要考虑各个类别的漏识别率、误识别率指标,需要对不同的错误指标设计对应的分类器,文中提出利用三个2种类型的分类器输出结果,进行组合的方式,获得能够满足多种分类误差条件的分类器设计方法,所提出的方法简化了分类器设计复杂度,通过分类器输出结果的映射变换,能够快速实现满足不同分类误差要求的分类。The research presents a classification method for persons＇ Gensini parameters range based on routine physical examination data, that distinguish which intervals of Gensini parameters the person involved. Because of determining the classification performance needs to consider the leakage recognition rate for each category, false identification rate index, it needs to design corresponding classifier for different error indicators. The paper proposed the use of three 2-types classifiers＇ output, to obtain a classifier design method can meet a variety of classification error conditions by combination. The proposed method simplifies the design complexity of classifier, through the mapping transformation of the classifiers＇ output, the classification can be achieved quickly to meet the requirements of different classification error.

关 键 词：Gensini参数 分类 体检数据 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]