阿尔茨海默病诊断的自适应BP神经网络模型

Adaptive BP Neural Network Model for Diagnosing Alzheimer's Disease

阿尔茨海默病(AD)和轻度认知功能损伤(MCI)具有患者多、诊断难的特点,改进BP神经网络,提出自适应BP神经网络(ABP)进行100次AD和MCI诊断模拟,ABP神经网络的诊断正确率显著高于BP和RBF神经网络.采用留一法将101例正常人、200例MCI和90例AD患者的样本分为训练集和检测集,用ABP神经网络对其进行诊断模拟,总正确率达到73.91%.

There are many Alzheimer＇s disease（AD） and mild cognitive impairment（MCI）patients in the world,who are difficult to diagnose.Improve BP neural network and put forward an adaptive BP neural network（ABP） to make 100 times of AD and MCI diagnosis simulations.The diagnostic accuracy of ABP neural network is significantly higher than those of BP and RBF neural network.Finally,all samples including 101 normal persons,200 MCI patients and 90 AD patients were divided into training set and checking set according to leave-one-out method and made diagnosis simulation,whose total accuracy reaches 73.91%.