神经网络模型在纵向随访数据分析中的研究进展  

Research progress on neural network model in longitudinal follow-up data analysis

作　　者：张文茜[1] 苏海霞[2] 孙丽君[1] 张玉海[1] 

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

出　　处：《实用预防医学》2017年第1期127-128,F0003,共3页Practical Preventive Medicine

基　　金：国家自然科学基金(81573252)

摘　　要：流行病学随访研究中会产生大量的纵向数据,但该类数据的分析一直是统计学的难点。目前大多采用传统的线性混合模型来处理。该模型对数据的分布要求严格,且假定疾病是线性进展的,在应用过程中会受到一定限制。近年来,有研究者提出采用神经网络模型来处理纵向随访数据。本文就神经网络模型在纵向随访数据分析中的研究现状进行探讨,为纵向随访数据的分析提供一个新思路。Longitudinal data are generated in the epidemiologieal follow-up study, but the analysis of such data is always a difficult point in statistics. At present, we mostly use the traditional linear mixed model to analyze such longitudinal data. The model is strict in the distribution of data, moreover, it is assumed that the disease is a linear progression, its application would be subject to some restrictions. Some researchers have proposed using the neural network model to analyze longitudinal follow-up data in recent years. This review discusses the current research status of neural network model in longitudinal follow-up data analysis and provides a new way for the analysis of longitudinal follow-up data.

关 键 词：纵向数据 线性混合模型 神经网络模型 

分 类 号：R195.1[医药卫生—卫生统计学]