职业与健康 ›› 2026, Vol. 42 ›› Issue (14): 1952-1958.

• 论著—流行病学与预防保健 • 上一篇    下一篇

基于随机森林的山东农村老年人抑郁症状影响因素分析

丁冉1, 刘啸宇1, 冷敏敏2, 杨丽娟2, 张爱华1()   

  1. 1 山东第一医科大学(山东省医学科学院)山东 泰安 271016
    2 山东第一医科大学附属省立医院山东 济南 250021
  • 收稿日期:2025-04-07 修回日期:2025-09-05 出版日期:2026-07-15 发布日期:2026-07-16
  • 通信作者: 张爱华,E-mail:zhangah818@163.com
  • 作者简介:丁冉,女,在读硕士研究生,研究方向为护理心理。
  • 基金资助:
    中国博士后科学基金(2023M732119);中国博士后科学基金(2023TQ0203);山东省自然基金(ZR2024QG032)

Analysis of the influencing factors of depressive symptoms among rural elderly people in Shandong Province based on the random forest algorithm

DING Ran1, LIU Xiaoyu1, LENG Minmin2, YANG Lijuan2, ZHANG Aihua1()   

  1. 1 Shandong First Medical University&Shandong Academy of Medical SciencesTaianShandong 271016, China
    2 Shandong Provincial Hospital Affiliated to Shandong First Medical UniversityJinanShandong 250021, China
  • Received:2025-04-07 Revised:2025-09-05 Online:2026-07-15 Published:2026-07-16
  • Contact: ZHANG Aihua,E-mail:zhangah818@163.com

摘要:

目的 基于随机森林算法分析山东省农村老年人抑郁症状影响因素,为防控老年人抑郁症状的发展提供依据。方法 采用便利抽样法,2024年4—6月选取5 136名山东省农村老年人,借助患者健康状况问卷筛选抑郁症状。使用随机森林算法筛选抑郁症状影响因素并排序,通过多因素Logistic回归分析其影响因素。结果 5 136名农村老年人的抑郁症状得分[MP25P75)]为1.00(0.00,4.00)分,总体检出率为19.5%。随机森林算法显示,睡眠质量、自评健康状况和医疗支付方式为重要性排序前3位的变量。Logistic回归分析结果显示,年龄≥80岁(OR=0.740,95%CI:0.556~0.985)、月收入1 000~<3 500元(OR=0.704,95%CI:0.525~0.945)、月收入≥3 500元(OR=0.555,95%CI:0.341~0.903)、医疗支付方式为居民医保(OR=0.477,95%CI:0.351~0.649)、饮酒(OR=0.757,95%CI:0.600~0.954)、睡眠质量好(OR=0.190,95%CI:0.160~0.225)、中等水平社会支持(OR=0.319,95%CI:0.170~0.601)和高水平社会支持(OR=0.276,95%CI:0.144~0.529)均是老年人抑郁症状的保护因素(均P<0.05);服药种类≥4种(OR=1.527,95%CI:1.147~2.033)、慢性病数量3~6种(OR=1.709,95%CI:1.176~2.483)、慢性病数量≥7种(OR=2.022,95%CI:1.245~3.283)、视力下降(OR=1.492,95%CI:1.261~1.765)、自评健康状况一般(OR=3.875,95%CI:2.916~5.151)和自评健康状况差(OR=5.295,95%CI:2.55~10.985))、日常生活活动能力下降(OR=1.894,95%CI:1.380~2.599)及认知功能下降(OR=1.371,95%CI:1.121~1.678)均是老年人抑郁症状的危险因素(均P<0.05)。结论 山东省5 136名农村老年人的抑郁症状检出率较高,影响因素较多,年龄、月收入、医疗支付方式、用药种类、饮酒、慢性病数量、视力、自评健康状况、认知功能、睡眠质量、日常生活活动能力及社会支持水平均是其影响因素。应从各个层面,针对重点人群,联合采取有效干预措施,减少山东省农村老年人群抑郁症状的发生。

关键词: 随机森林, 农村, 老年人, 抑郁症状, 影响因素

Abstract:

Objective To analyze the current situation and influencing factors of depressive symptoms among rural elderly people in Shandong Province based on the random forest algorithm,aiming to provide a basis for preventing and controlling the occurrence of depressive symptoms in the elderly. Methods Convenience sampling was used to select 5 136 rural elderly people in Shandong Province from April to June 2024,and the patient health questionnaire was used to screen for depressive symptoms.The random forest algorithm was used to screen and rank the influencing factors of depressive symptoms,and multi-factor Logistic regression was used to analyze these influencing factors. Results The score of depressive symptoms[MP25P75)] among 5 136 rural elderly people was 1.00(0.00,4.00)points,and the overall detection rate was 19.5%.The random forest algorithm showed that sleep quality,self-rated health status,and medical payment method were the top three variables in terms of importance. The results of Logistic regression analysis showed that age≥80 years(OR=0.740,95%CI:0.556-0.985),monthly income of 1 000-<3 500 yuan(OR=0.704,95%CI:0.525-0.945),monthly income ≥3 500 yuan(OR=0.555,95%CI:0.341-0.903),medical payment method being resident medical insurance(OR=0.477,95%CI:0.351-0.649),alcohol consumption(OR=0.757,95%CI:0.600-0.954),good sleep quality(OR=0.190,95%CI:0.160-0.225),moderate social support level(OR=0.319,95%CI:0.170-0.601),and high social support level(OR=0.276,95%CI:0.144-0.529) were protective factors for depressive symptoms in the elderly(all P<0.05). ≥4 types of medication(OR=1.527,95%CI:1.147-2.033),3-6 types of chronic diseases(OR=1.709,95%CI:1.176-2.483),≥7 types of chronic diseases(OR=2.022,95%CI:1.245-3.283),decreased vision(OR=1.492,95%CI=1.261-1.765),self-rated general health status(OR=3.875,95%CI:2.916-5.151) and self-rated poor health status(OR=5.295,95%CI:2.55-10.985),decreased ability in daily living activities(OR=1.894,95%CI:1.380-2.599),and decline in cognitive function(OR=1.371,95%CI:1.121-1.678) were risk factors for depressive symptoms in the elderly(all P<0.05). Conclusion The incidence of depressive symptoms among the elderly in rural areas in Shandong Province is relatively high,and there are many influencing factors,including age,monthly income,medical payment method,types of medication,alcohol consumption,number of chronic diseases,vision,self-rated health status,cognitive function,sleep quality,activities of daily living,and social support level.Effective intervention measures should be jointly taken at all levels and targeted at key populations to reduce the occurrence of depressive symptoms among rural elderly people in Shandong.

Key words: Random forest algorithm, Rural area, Elderly, Depressive symptoms, Influencing factors

中图分类号: