基于P-trees kNN算法的毒物分类方法  

A Poison Classification Diagnosis Method Based on P-trees KNN

作　　者：曾志浩[1] 胡积平[2] 

机构地区：[1]湖南工业大学计算机与通信学院 [2]国防科学技术大学计算机学院

出　　处：《软件》2012年第4期105-107,111,共4页Software

基　　金：湖南省教育厅科学研究项目(11C0400)

摘　　要：中毒是一种发生机率较大、对人体危害大的病症,而及时明确诊断,正确、规范的治疗既是抢救成功的关键,又是至今没有很好解决的难题。毒物层出不穷,基层急救医生的毒物知识和中毒抢救知识又明显不足,因此临床急需一种辅助系统以帮助各级急救医生提高中毒诊治水平及中毒抢救成功率。利用不同中毒表现对应不同毒物的权值向量构成"中毒表现加权向量表",并将它作为训练数据集的属性值。构建中毒表现加权向量表的P树,并选择HOBBit距离作为距离度量标准,运用P-trees kNN分类算法进行毒物分类。将该方法应用到毒物分类系统中,运行效果良好。Poisoning is a kind of disease of greater probability takes place and endangers to the human body. Diagnose clearly in time, correct, normal treatment is the key to rescuing successfully which is the difficult problem have not be solved. The poisonous sub- stance emerges in an endless stream, the basic level doctor＇s poisonous substance knowledge and poisoning rescue knowledge is obviously insufficient. So clinical need some accessory system badly to help doctors approving their poison diagnosis treatment and poison rescuing success rate. By statistic it can obtain poison＇s weighted vectors corresponding difference clinical symptoms. Thus can get weighted vector symptom table from these vectors and treats is as the attribute value of training dataset. With establishing the P-tree using these weighted vectors according the HOBBit distance measure, it can rapidly give possible poisoning diagnosis.

关 键 词：K近邻算法 中毒分析系统 中毒表现加权向量表 P-树 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]