支持向量机算法研究头发微量元素与高血压的相关性  

Support vector machine applied to the relationship between the content of some trace elements in hair and hypertension disease

作　　者：陈瑞兰[1] 陆文聪[1] 刘旭[1] 陈念贻[1] 

机构地区：[1]上海大学化学系计算机化学研究室,上海200436

出　　处：《计算机与应用化学》2003年第5期567-570,共4页Computers and Applied Chemistry

基　　金：国家自然科学基金(20006008)

摘　　要：将Vapnik提出的支持向量机(Support Vector Machine,SVM)算法用于总结头发中多种微量元素含量与高血压的对应关系的结果。通过对26个高血压患者和27个健康人的头发样品的多种微量元素的定量分析,用支持向量机研究头发微量元素与高血压的相关性,结果表明:若以头发中Al,Cu,Zn,Ca,Mg含量以及Zn/Cu比作为特征量集合作数据挖掘,所建数学模型对高血压患者与健康人的正确分类率可达96.2％,留一法预报正确率则可达86.7％。计算表明:支持向量机算法建模的正确分类率和留一法预报正确率均较Fisher法和KNN法等传统的模式识别算法高。因此,SVM算法是特别适合于用有限已知样本训练建模,进而预报未知样本属性的新算法,并可望在化学计量学领域得到进一步的应用。The trace element content of the hair samples of 26 high blood pressure patients and 27 healthy people have been determined. And the data of Al, Ca, Cu, Zn, Mg contents have been processed by a new data mining algorithm, support vector machine proposed by Vapnik, in order to search the correlation between these data and high blood pressure disease. It has been found that the rate of correct classification is 96. 2% , and the rate of correctness of prediction by LOO method is 86. 7% . The results of SVM appears somewhat better than that obtained by some traditional classification methods such as Fisher method or KNN. Therefore, support vector machine is suitable for the data processing based on finite number of training samples, with special technique to restrict overfitting. It is expecled that the SVM melhod would be further applied lo the field of chemometric studies.

关 键 词：支持向量机算法 头发 微量元素 高血压 相关性 数学模型 

分 类 号：R544.1[医药卫生—心血管疾病] R446.19[医药卫生—内科学]