基于BP神经网络和RBF神经网络预测老年痴呆症疾病进展的对比研究  

A Comparative Study on Prediction of Alzheimer's Disease Progression based on BP and RBF Neural Network

作　　者：张文茜[1] 苏海霞[2] 尚磊[1] 孙丽君[1] 张玉海[1] 

机构地区：[1]第四军医大学卫生统计教研室,陕西西安710032 [2]第四军医大学流行病学教研室,陕西西安710032

出　　处：《现代生物医学进展》2017年第4期738-741,共4页Progress in Modern Biomedicine

基　　金：国家自然科学基金项目(81573252)

摘　　要：目的:比较反向传播算法(BP)神经网络和径向基函数(RBF)神经网络预测老年痴呆症疾病进展的效果。方法:以老年痴呆症随访数据为研究对象,以性别、年龄、受教育程度、有无高血压、有无高胆固醇、有无心脏病、有无中风史、有无家族史8个指标作为输入变量,以五年随访的MMSE差值为输出变量,构建基于BP神经网络和RBF神经网络的老年痴呆症疾病进展预测模型。结果:与BP神经网络模型相比,RBF神经网络预测的结果更好,能够有效地预测老年痴呆症疾病进展。结论:神经网络模型将老年痴呆症疾病进展预测问题转化为随访数据中相关测量指标与MMSE差值的非线性问题,为复杂的老年痴呆症疾病进展预测提供了新思路。Objective： To compare the effects of BP and RBF neural network for predicting Alzheimer＇s disease progression.Methods： Gender, age, education level, presence versus absence of hypertension, hypercholesterolemia, heart disease, stroke, and family dementia history were selected as input variables, the MMSE difference of five years follow-up was selected as the output variable for BP and RBF Neural networks prediction models. Results： Compared with the BP neural network model, RBF neural network prediction results was better and can effectively predict the progression of Alzheimer＇s disease. Conclusions： Neural network models transform the Alzheimer＇s disease progression prediction into the nonlinear problem of follow-up data on relevant measurement index and MMSE difference, which provides a new idea for the prediction of the complex Alzheimer＇s disease progression.

关 键 词：老年痴呆症 反向传播算法神经网络 径向基函数神经网络 预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]