基于正则化逻辑回归的阿尔茨海默病早期诊断模型  

Early diagnosis model of Alzheimer’s disease based on regularized logistic regression

作　　者：张贻泉 徐培然 韩莉莉[3] 张宝昌[4] 肖如意 刘汉磊 崔新春[1] ZHANG Yiquan;XU Peiran;HAN Lili;ZHANG Baochang;XIAO Ruyi;LIU Hanlei;CUI Xinchun(School of Computer Science,Qufu Normal University,276826,Rizhao;Department of Computer Science,Jining University,272071,Jining;School of Management,Qufu Normal University,276826,Rizhao;Center of Network and Information,Jining Medical University,273100,Jining,Shandong,PRC)

机构地区：[1]曲阜师范大学计算机学院,日照市276826 [2]济宁学院计算机科学系,济宁市272071 [3]曲阜师范大学管理学院,日照市276826 [4]济宁医学院网络信息中心,山东省济宁市273100

出　　处：《曲阜师范大学学报（自然科学版）》2021年第4期65-71,共7页Journal of Qufu Normal University(Natural Science)

基　　金：教育部人文社会科学项目(11yjczh021,15yjczh111);山东省社会科学规划研究项目(17chlj41,16ctqj02,18chlj34);山东省社科规划项目数字山东研究专项(20CSDJ20);济宁医学院教师科研扶持基金.

摘　　要：提出了一种基于L_(2)正则化逻辑回归(LR)的阿尔茨海默病(AD)诊断算法.在该模型中使用了L_(2)范数对LR进行正则化处理,正则化参数通过十倍交叉验证来选择,同时使用了独立成分分析对预处理后的数据进行降维处理,最后使用了牛顿算法来求出模型的最优权值.通过这一算法可以有效分辨出AD及其早期阶段轻度认知障碍(MCI).实验在AD vs.CN,MCI vs.CN和LMCI vs.EMCI 3组分类任务中获得的分类准确率分别为95.22%,81.22%和74.35%.实验结果证明其为一种有效的诊断算法.In this paper,an algorithm for the diagnosis of Alzheimer's disease(AD)based on L_(2) regularized logistic regression(LR)is proposed.Among them,the L_(2) norm is used to regularize the LR.The regularization parameters are selected through ten-fold cross-validation.At the same time,independent component analysis is used to reduce the dimensionality of the preprocessed data.Finally,the Newton algorithm is used to find the optimal weight of the model.This algorithm can effectively distinguish AD and its early-stage mild cognitive impairment(MCI).The classification accuracies obtained by the experiment in the three groups of classification tasks of AD vs.CN,MCI vs.CN and LMCI vs.EMCI were 95.22%,81.22% and 74.35%,respectively.The results proved that it is an effective diagnostic algorithm.

关 键 词：阿尔茨海默病 轻度认知障碍 正则化逻辑回归 独立成分分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]