计算机技术与发展2018，Vol.28Issue(3)：178-182,5.DOI:10.3969/j.issn.1673-629X.2018.03.038
M3-SVM在帕金森疾病UPDRS分类中的应用
Application of Min-Max Modular SVM in UPDRS Classification of Parkinson's Disease
摘要
Abstract
Based on the fact that the effect of Parkinson's disease on the phonetics of male and female patients is different,we propose in-corporating the prior knowledge of gender into the min-max modular support vector machine(M3-SVM) to achieve the classification of unified Parkinson's rating scale(UPDRS).Firstly,the task is decomposed by using gender as the prior knowledge.Secondly,the training subsets are trained by support vector machine(SVM) to get the base classifier.Finally,the MIN and MAX rules are used to integrate the results of the classifier for the final classification results.In the simulation experiments of two and three Parkinsons Telemonitoring Data Set,the best F-measure values are 80.19% and 69.26% by gender partition respectively,which are 3.75% and 5.19% higher than that of random partition,and 0.96% and 4.15% higher than that of hyperplane partition.The experiments show that M3-SVM based on gen-der partition can improve the accuracy of UPDRS classification.关键词
帕金森疾病/语音/最小最大模块化支持向量机/性别划分/统一帕金森评定量表Key words
Parkinson's disease/speech/min-max modular support vector machine/gender partition/unified Parkinson's disease rating scale分类
信息技术与安全科学引用本文复制引用
汪学明,季薇,李云..M3-SVM在帕金森疾病UPDRS分类中的应用[J].计算机技术与发展,2018,28(3):178-182,5.基金项目
国家自然科学基金(61603197) （61603197）
江苏省自然科学基金(BK20140885) （BK20140885）
江苏省博士后基金(1401045C) （1401045C）
南京邮电大学科研基金(NY214034,NY215104,NY214127) （NY214034,NY215104,NY214127）
