摘要
老年人自评健康是反映老年人身体健康状态的重要因子,对提高老年人健康水平提供参考具有重要意义。为了解影响我国农村老年人自评健康的主要因素并实现精准地预测,本研究基于2022年湖南省岳阳县养老需求调研数据,首先探究了不同影响因素对老年人自评健康的作用机制;然后基于显著影响因素,在面向高维度数据特征的情况下,提出一种基于交叉熵和变学习率的改进一维卷积神经网络(1D-ICNN)用于构建老年人自评健康预测模型,以解决1D-CNN容易出现预测不准确和不稳定等问题。本研究显示,老年人自评健康与文化程度、政治面貌、婚姻状况、职业、年收入等因素有关;在较高维度数据特征情况下,1D-ICNN模型具有较好的预测效果。该方法的应用和普及能够为准确预测老年人健康状况、实现“健康老龄化”提供实证依据。
The self-rated health of the elderly is an important factor to reflect the health status of the elderly,and it is of great significance to provide reference for improving the health level of the elderly.In order to understand the main factors affecting the self-rated health of the rural elderly in China and achieve accurate prediction,this study first explored the mechanism of different influencing factors on the self-rated health of the elderly based on the survey data of the elderly care demand in Yueyang County,Hunan Province in 2022.Then,based on the significant influencing factors,an improved one-dimensional convolutional neural network(1D-ICNN)based on cross entropy and variable learning rate is proposed to construct a self-rated health prediction model for the elderly in the case of high-dimensional data features,so as to solve the problems of inaccurate prediction and instability of 1D-CNN.This study shows that the self-rated health of the elderly is related to factors such as education level,political outlook,marital status,occupation and annual income.In the case of higher dimensional data features,the 1D-ICNN model has better prediction results.The application and popularization of this method can provide an empirical basis for accurately predicting the health status of the elderly and achieving"healthy aging".
作者
李玥
张承蒙
黄成烨
索浩宇
胡新悦
刘娜
张雅璐
陈功
LI Yue;ZHANG Cheng-meng;HUANG Cheng-ye;SUO Hao-yu;HU Xin-yue;LIU Na;ZHANG Ya-lu;CHEN Gong(Institute of Population Research,Peking University,Beijing 100871,China)
出处
《医学信息》
2024年第14期25-32,共8页
Journal of Medical Information
基金
中国工程院战略研究与咨询项目(编号:2022-XBZD-30)
国家社会科学基金青年项目(编号:22CRK005)。
关键词
老年人
自评健康
一维卷积神经网络
预测模型
Elderly
Self-rated health
One-dimensional convolutional neural network
Prediction model
