摘要
将支持向量机(SVM)应用于老年痴呆症(AD)的模式识别研究。通过测定22个AD患者和25个健康人头发样品中微量元素的含量,继用支持向量机算法研究头发中微量元素含量与AD的相关性,建立分类判别模型。结果显示:该模型对AD的判别准确率为100%,留一法交互预测准确率也达到100%。变量筛选结果表明,与AD症相关性最大的三种元素是Al、Cd、Mn,Al、Cd与AD呈现正相关,Mn与AD呈现负相关。同时与主成分分析进行了比较,表明SVM是更适合于进行这类非线性多变量相关分析的方法。
Support Vector Machine was applied to study the recognition models of Alzheimer＇s disease. The relationships between the contentof some trace elements in hair and Alzheimer＇s disease haven been studied by the determinations of the trace element content of the hair samples of 22 AD patients and 25 heathy people, and the classfication recognition models also been constructed. It has been found that the rate of correct classification is 100%, and the rate of correctness of prediction by Leave One Out method is also 100%. The results of variable selection implied that the element ofAl, Cd, Mn have the great relationshipments with AD, and AI, Cd element have the positive correlation, Mn element has the negative corrlation. The results implied SVM is more suitable for this class of nonlinear multivariable correlation analysis method compare to principle component analysis.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2013年第2期125-128,共4页
Computers and Applied Chemistry
基金
湖南省高校"中药制剂工程与质量控制"产学研合作示范基地(2010)资助
民族药用植物资源研究与利用湖南省重点实验室资助项目(HHUW2011-68)
作者简介
杨兴华（1961—），男，湖南人，教授，Email：．hhyangxh@163．com
