广义回归神经网络在阿尔茨海默病诊断中的应用  

Applications in the diagnosis of Alzheimer′s disease with generalized regression neural network

作　　者：罗万春 马翠 宋丽娟 魏调霞 LUO Wanchun;MA Cui;SONG Lijuan;WEI Tiaoxia(Department of Mathematics,College of Basic Medicine,Army Medical University,Chongqing 400038,China)

机构地区：[1]陆军军医大学基础医学院数学教研室,重庆400038

出　　处：《现代医药卫生》2022年第15期2527-2530,共4页Journal of Modern Medicine & Health

基　　金：重庆市自然科学基金项目资助(CSTC2013jcyjA10041)。

摘　　要：目的建立正确率较高的阿尔茨海默病(AD)和轻度认知功能损伤(MCI)的诊断数学模型。方法以筛选出的4项指标为输入变量,将391个样本随机分组为训练集和检测集,用广义回归神经网络(GRNN)在训练集和检测集中分别进行参数训练和模拟诊断,并与BP神经网络(BPNN)、径向基神经网络(RBFNN)和感知器神经网络(PNN)进行诊断效果比较。结果GRNN对于3类人群的诊断正确率显著高于BPNN、RBFNN和PNN,差异均有统计学意义(P<0.05)。当GRNN取最优平滑因子1.5时,391个样本的诊断正确率达到75.7%。结论GRNN模型对AD和MCI的诊断效果较好,可以作为一种临床诊断辅助手段。Objective To establish a mathematical diagnostic model of Alzheimer′s disease(AD)and mild cognitive impairment(MCI)with a higher accuracy.Methods Taking the four selected indexes as input variables,391 samples were randomly divided into the training set and the testing set.Generalized regression neural network(GRNN)was used for parameter training and simulation diagnosis in the training set and the testing set respectively,and the diagnosis effects were compared with those of BP neural network(BPNN),radial basis function neural network(RBFNN)and perceptron neural network(PNN).Results The diagnostic accuracy rate of GRNN for the three groups of people was significantly higher than that of BPNN,RBFNN and PNN,the differences were statistically significant(P<0.05).When the optimal smoothing factor of GRNN was 1.5,the diagnostic accuracy rate of 391 samples reached 75.7%.Conclusion The GRNN model has good diagnostic effect on AD and MCI,which can be used as an auxiliary means of clinical diagnosis.

关 键 词：阿尔茨海默病 轻度认知功能损伤 诊断 广义回归神经网络 

分 类 号：R311[医药卫生—基础医学]