基于MLP神经网络的自评健康和ADL对老年人抑郁的影响

The Impact of Self-Rated Health and Activities of Daily Living on Depression in the Elderly Based on MLP Neural Networks

摘要: 目的：研究自评健康和日常生活活动能力(ADL)对老年人抑郁的影响，为其防治提供数据支撑。方法：利用2018年中国老年健康数据调查的数据，以9969例老年人为研究对象，抑郁的测量采用流调中心抑郁量表。利用卡方检验和二元Logistic回归分析分析影响老年人抑郁的因素。构建多层感知器(MLP)的神经网络，探究自评健康和ADL对老年人抑郁的影响的重要性。结果：二元Logistic回归分析的结果显示：女性(OR = 1.119, 95% CI = 1.031~1.214)、农村(OR = 1.380, 95% CI = 1.260~1.511)、≥85岁(OR = 1.311, 95% CI = 1.199~1.433)、ADL失能(OR = 1.283, 95% CI = 1.145~1.439)、独居(OR = 1.311, 95% CI = 1.174~1.465)、居住在养老机构(OR = 1.544, 95% CI = 1.228~1.940)、自评健康一般(OR = 1.492, 95% CI = 1.367~1.629)、自评健康不好(OR = 2.728, 95% CI = 2.386~3.118)、经济状况一般(OR = 1.144, 95% CI = 1.030~1.270)及经济状况困难(OR = 1.723, 95% CI = 1.457~2.037)会增加老年人抑郁的风险。MLP神经网络的重要性排序：自评健康(0.258)、ADL(0.233)、经济状况(0.188)、年龄(0.156)、居住方式(0.088)、民族(0.035)、城乡分布(0.027)、性别(0.026)。结论：自评健康和ADL都会影响老年人抑郁，且自评健康的重要性大于ADL的重要性。改善老年人自评健康状况，鼓励ADL失能的老年人进行康复训练并经常安排心理疏导，有助于降低老年人发生抑郁的风险。

Abstract: Objective: To study the effects of self-rated health and activities of daily living on depression in the elderly, and to provide data support for its Prevention and treatment. Methods: Based on the data of the 2018 China Elderly Health Data Survey, 9969 elderly people were selected as the research objects, and depression was measured using center for epidemiologic studies depression scale. Chi-square test and binary logistic regression analysis were used to analyze the factors influencing depression in the elderly. The importance of constructing a neural network with multilayer perceptron to explore the impact of self-rated health and daily living ability on depression in older adults. Results: The results of binary logistic regression analysis showed that female (OR = 1.119, 95% CI = 1.031~1.214), rural (OR = 1.380, 95% CI = 1.260~1.511), ≥85 years old (OR = 1.311, 95% CI = 1.199~1.433), ADL disability (OR = 1.283, 95% CI = 1.145~1.439), and living alone (OR = 1.311, 95% CI = 1.174~1.465), living in nursing institutions (OR = 1.544, 95% CI = 1.228~1.940), self-rated average healthy (OR = 1.492, 95% CI = 1.367~1.629), self-rated poor healthy (OR = 2.728, 95% CI = 2.386~3.118), average economic status (OR = 1.144, 95% CI = 1.030~1.270), and difficult economic status (OR = 1.723, 95% CI = 1.457~2.037) increased the risk of depression in older adults. The importance ranking of MLP neural networks was self-rated health (0.258), ADL (0.233), economic status (0.188), age (0.156), residence style (0.088), ethnicity (0.035), urban-rural distribution (0.027), and gender (0.026). Conclusion: Both self-rated health and ability to perform activities of daily living affect depression in older adults, and the importance of self-rated health is greater than that of activities of daily living. Improving the self-rated health status of older adults, encouraging rehabilitation training for older adults with ADL disability, and arranging psychological counseling frequently can help reduce the risk of depression in older adults.