PSYCH OpenIR  > 健康与遗传心理学研究室
基于随机森林算法的青少年心理健康测量和影响因素分析
其他题名Analysis of Important Items and Factors of Adolescent Mental Health: A Random Forest Approach
潘晓君
导师陈祉妍
2023-12
摘要根据联合国儿童基金(UNICEF)的统计,全球每 7 个 10-19 岁的青少年中就有 1 个有心理健康问题,约占青少年总人数的 14%。青少年心理健康问题不仅影响青少年当前的生活,而且会对以后的成年生活持续造成负面影响。因此,国家卫建委等 12 个部门呼吁,“针对儿童青少年常见的心理行为问题与精神障碍,开展早期识别与干预研究”。快速检测青少年的心理问题,并准确甄别其主要影响因素,是青少年心理健康问题早发现、早干预的基石。以往的研究已经提供了成熟的量表测量青少年的心理健康水平,并且局部探索了多种青少年心理健康的影响因素。本研究着眼于青少年的心理健康问题的早期识别、提前预防、尽早干预, 拟从常用量表中甄别出测量青少年心理健康最重要的条目,并从大量潜在影响因素中识别与青少年心理健康水平关系更加密切的影响因素。 本研究数据来源于国民心理健康数据库(Chinese Mental Health Database, CMHD)中的全国学生心理健康状况调查。该调查覆盖31个省(直辖市、自治区),包含 192 所小学、96 所初中、96 所高中和 128 所高校。采用分层取样,每所学校有 2 个班级的所有学生参与了调查,共获得 68191 份有效问卷。本研究采用机器学习随机森林算法,进行心理健康测量条目和影响因素的重要性分析:研究一基于心理健康两因素模型,甄别采用流调中心抑郁量表、广泛性焦虑障碍量表和简版多维生活满意度量表筛查青少年心理健康水平的最重要条目;研究二基于生态系统理论,探索与青少年心理健康水平更加密切相关的因素。 研究一测量条目重要性分析结果显示:(1)流调中心抑郁量表的5个消极情 绪维度条目是测量抑郁的重要条目,模型准确度为 90%;(2)广泛性焦虑障碍量 表的2个消极认知和情绪维度条目是测量焦虑的重要条目,模型准确度为95%; (3)简版多维生活满意度量表的总体生活满意度和自我满意度是测量青少年生活满意度的重要条目,模型准确度为 87%;(4)青少年心理健康四水平(完全健康组、部分健康组、部分病态组和完全病态)的 10 个重要测量条目包括简版多维生活满意度量表和流调中心抑郁量表的消极情绪维度条目,模型准确度为 88%。 研究二影响因素重要性分析结果显示:学段以及与学校微系统相关的影响 因素(学校环境、同伴关系、同伴欺凌受害、师生关系、学习兴趣)与青少年心 理健康水平更加密切相关。其中,(1)与抑郁关系最密切的 5 个因素依次是学校环境、同伴关系、同伴欺凌受害、年级和师生关系;(2)与焦虑关系最密切的 5 个因素因此是学校环境、校园霸凌、年级、师生关系和同伴关系;(3)与生活满意度最密切的5个因素依次是同伴关系、学校环境、学习兴趣、父子关系和 年级; (4) 与4 种青少年心理健康水平最密切相关的5个因素依次是学校环境、 同伴关系、师生关系、年级和同伴欺凌受害。 综上所述,本研究提供了快速检测青少年心理问题的方法,并且识别了与青少年心理健康关系最密切的因素。基于以上研究结果,抑郁和焦虑的主要测量条目是消极情绪,因此日常生活中父母和学校要关注青少年的消极情绪,以便早日发现青少年的心理健康问题。同时,针对不同心理健康类型的青少年,建议采用三级防御预防措施以及干预治疗,构建安全友爱和谐的校园秩序,杜绝校园霸凌,通过增强青少年情绪调节能力、校园归属感以及应对学业压力的能力,最终提升青少年的心理健康水平。
其他摘要According to the statistical data from UNICEF, it is estimated that 1 in 7 (14%) 10-19-year-olds experience mental disorders globally. Adolescent mental disorders not only affect their life in the short term but can also continue to impact them during their adult years negatively. Therefore, twelve government departments, including the National Health Commission of the People’s Republic of China, have called for studies on early detection and intervention of children and adolescents with common psychobehavioral and mental disorders. Early detection and intervention of adolescent mental disorders are facilitated by the approach of rapid detection of adolescent mental disorders and the screening of major influencing factors. Previous studies have provided well-established scales to measure adolescent mental health and have partially explored a variety of influence factors on adolescent mental health. This study focuses on the early identification, prevention, and intervention of mental health problems in adolescents, and aims to identify the most important items for measuring adolescents' mental health from commonly-used scales, as well as to identify influential factors that are more closely related to the level of adolescents' mental health from a large number of potential influential factors. The data for this study came from the National Student Mental Health Survey in the Chinese Mental Health Database. The survey covered 31 provinces (municipalities and autonomous regions) and included 192 elementary schools, 96 middle schools, 96 high schools, and 128 colleges and universities. Stratified sampling was used and all students in 2 classes in each school participated in the survey and a total of 68,191 valid questionnaires were obtained. Based on the big data above, this study conducted an importance analysis of mental health measurement entries and influencing factors using the machine learning algorithm random forest. Based on the Dual-Factor Model of Mental Health, study 1 explored the rapid screening of adolescent mental health status using The Center for Epidemiological Studies Depression Scale (CESD-20), Generalized Anxiety Disorder Scale (GAD-7), and Brief Multidimensional Students' Life Satisfaction Scale (BMSLSS). Based on the Dual-Factor Model of Mental Health and the Bioecological Model, study 2 examined environmental factors more closely related to adolescent mental health. The results of the importance analysis of the measurement entries in Study 1 showed that, (1) the five negative affection dimension entries of CESD-20 were the most important entries for measuring depression with a model accuracy of 90%; (2) the two negative cognitive and affective dimension entries of the GAD-7 were the most important entries for measuring anxiety with a model accuracy of 95%; (3) the overall life satisfaction entry and self-satisfaction entry in BMSLSS are the most important entries for measuring adolescent life satisfaction, with a model accuracy of 87%; (4) the 10 important measurements of adolescent mental health levels (Flourishing, Vulnerable, Symptomatic but content, Troubled) include all the entries in BMSLSS and the negative affection dimensions of CESD-20, with a model accuracy of 88%. The results of the importance analysis of the influencing factors in Study 2 showed that grade and the influencing factors related to the school microsystem (school environment, peer relationships, school bullying, teacher-student relationships, and interest in learning) were more closely related to adolescent mental health levels. Of these, (1) the five factors are most closely related to adolescent depression: school climate, peer friendship, school bullying, grade, and teacher-student relationship; (2) the five factors are most closely related to adolescent anxiety: school climate, school bullying, grade, teacher-student relationship, and peer friendship; (3) the five factors are most closely related to adolescent life satisfaction: peer friendship, school climate, academic interest, father-student relationships, and grade; and (4) the five factors are most closely associated with four adolescent mental health levels: school climate, peer friendship, teacher-student relationships, grade, and school bullying. In summary, this study provides a rapid method to detect adolescent mental problems and identifies the factors most closely related to adolescent mental health. Based on these findings, the main measurement entry for depression and anxiety is negative affection, so it is important for parents and schools to pay attention to adolescent negative moods in daily life in order to detect adolescent mental health problems as early as possible. Meanwhile, it is recommended to use tertiary defense preventive measures as well as intervention treatments for adolescents with different mental health levels by building a safe, friendly, and harmonious school climate, eliminating school bullying, enhancing adolescent emotion regulation skill, school belongness and the ability to cope with academic pressure.
关键词青少年心理健康 心理健康双因素模型 生态系统理论 随机森林
学位类型继续教育硕士
语种中文
学位名称理学硕士
学位专业健康心理学
学位授予单位中国科学院大学
学位授予地点中国科学院心理研究所
文献类型学位论文
条目标识符http://ir.psych.ac.cn/handle/311026/48222
专题健康与遗传心理学研究室
推荐引用方式
GB/T 7714
潘晓君. 基于随机森林算法的青少年心理健康测量和影响因素分析[D]. 中国科学院心理研究所. 中国科学院大学,2023.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
潘晓君-同等学力论文.pdf(1700KB)学位论文 限制开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[潘晓君]的文章
百度学术
百度学术中相似的文章
[潘晓君]的文章
必应学术
必应学术中相似的文章
[潘晓君]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。