- 无标题文档
查看论文信息

中文题名:

 基于生态系统理论和认知模型的心理健康情境测验及其应用    

姓名:

 王志蒙    

保密级别:

 公开    

论文语种:

 chi    

学科代码:

 045400    

学科专业:

 应用心理    

学生类型:

 硕士    

学位:

 应用心理硕士    

学位类型:

 专业学位    

学位年度:

 2023    

校区:

 北京校区培养    

学院:

 心理学部    

研究方向:

 用户体验    

第一导师姓名:

 蒋挺    

第一导师单位:

 心理学部    

提交日期:

 2023-06-25    

答辩日期:

 2023-05-22    

外文题名:

 A SITUATIONAL TEST OF MENTAL HEALTH BASED ON ECOSYSTEM THEORY AND COGNITIVE MODEL AND ITS APPLICATION    

中文关键词:

 心理健康 ; 抑郁 ; 学生 ; 情境测验 ; 大数据 ; 人工智能 ; ChatGPT ; ChatGLM-6B    

外文关键词:

 Mental Health ; Depression ; Students ; Situational Testing ; Big Data ; Artificial Intelligence ; ChatGPT ; ChatGLM-6B    

中文摘要:

青少年和青年学生心理健康问题是迄今仍未突破的重大公共卫生问题,需要突破现有 系统性理论不足和针对性服务缺乏等瓶颈,融合跨学科研究方法和技术,探索出一条契合 当代学生身心发展特征的理论创新与技术实践路径。为此,本研究开展了系列研究和实践: 
研究一基于生态系统理论和认知模型,融合行为事件访谈法、心理评估会谈法、网络 爬虫和大数据分析等研究方法和技术手段,梳理我国青少年和青年学生的不良心态类型。 182 名 12-24 岁中学生和大学生中筛选出 21 名处于轻、中和重度焦虑抑郁状态的受访者进 行长达 3-4.5 小时(60-90min/次 *3 次)的深度访谈。与此同时,从国内著名的心理咨询门 户网站——简单心理、壹心理、壹点灵等的心理问答模块成功爬取了 179 万余条心理问答 数据,经词频分析和人工筛选,纳入 807 条问题数据辅助质性分析。本研究以抑郁心态为 突破口,提出青少年和青年学生诱因型抑郁分类模型,包括学业成就型抑郁、身心倦怠型 抑郁、同伴关系型抑郁、家庭关系型抑郁、网络创伤型抑郁和身心暗示型抑郁等六大抑郁 类型,及其对应的典型情境、认知和行为特征。 
研究二在研究一的基础上研发具备生态效度的学生抑郁情境测验。经过研究者评估和 借助 ChatGPT(GPT-3.5 模型)反复润色和修订生成 38 项初版 12-24 岁抑郁情境测验。160 名参与者的有效数据被纳入预测试的数据分析;8 位来自心理测量、心理评估和心理咨询 等领域的资深专家评估了初版测验的有效性,涉及测验题目内容、选项内容和选项分数的 合理性、典型性和相关性等 684 个指标。结果表明,初版测验具有良好的内部一致性信度、 内容效度、校标关联效度,但结构效度较差。本研究进一步根据预测试和专家评估结果对 初版测验进行修订,最终生成 40 项正式版的抑郁情境测验。1275 名参与者的有效数据被 纳入数据分析,结果显示,该版本具备良好的内部一致性信度(Cronbach’s alpha = 0.932)、 校标关联效度(与贝克抑郁量表得分显著正相关,r=0.705,p<0.001;与患者健康问卷得分 显著正相关,r=0.659,p<0.001;与积极心理健康量表得分显著负相关,r=-0.565,p<0.001) 和结构效度(χ2=1482.239,χ2 /df=2.044,P<0.001,CFI=0.927,TLI=0.921,SRMR=0.035, Estimate=0.029),验证了从生态系统和认知模型角度共同评估青少年和青年学生抑郁状态 的可行性,为开发生态取向的心理健康评估工具奠定良好的实践基础。 
研究三基于前两个研究的成果,探索情境测验研究的数字化应用。基于人工智能和心 I理咨询技术,心理情境研究数据被用于 ChatGLM-6B 开源模型的微调,以初步研发 AI 心 理健康助手。微调后的模型被命名为心灵伴侣,6 名心理和计算机领域的工作者对其进行 可行性测试,验证其初步具备良好的训练效果,有望进一步生成心理健康和临床领域中的 研究型、培训型和陪伴型服务助手。 该系列研究成果为构建生态化、智能化和个性化的心理健康服务路径提供了可靠的理 论和实践基础,呼吁更多相关领域的研究者和实践人员从我国国情出发,深入了解不同人 群的日常心理需求,结合理论创新、技术融合和实践应用等多重驱力,共同完善中国特色 的社会心理服务体系,全面提升大众的心理健康和生活品质。

外文摘要:

Against the backdrop of constructing a social psychology service system with Chinese characteristics, the rapid growth of psychological problems among adolescents and young people has drawn extensive attention from all sectors of society. However, psychological health services for current adolescents and young people face a series of difficulties, such as a lack of systematic theory, insufficient effectiveness in practice, and incomplete targeted services. There is an urgent need to explore a theoretical and technological innovation path that fits the current characteristics of physical and mental development in adolescents and young people. For the aforementioned reasons, various studies have made efforts in this direction. 
Study1 based on the ecological system theory and cognitive model, integrating research methods and technical means such as behavioral event interviews, psychological assessment interviews, web crawlers, and big data analysis, to select 21 interviewees from 182 middle and high school students and university students aged 12-24 with mild, moderate, and severe anxiety and depression states. The interview process lasted for 3 to 4.5 hours (60 to 90 minutes per session). At the same time, more than 1.79 million pieces of psychological Q&A data were successfully crawled from well-known domestic psychology counseling platforms such as Jian Dan Xin Li, Yi Xin Li, and Yi Dian Ling. After word frequency analysis and manual screening, 807 pieces of question data were included to assist in qualitative analysis, jointly sorting out the types of negative attitudes among Chinese adolescents and young people. With depression as the breakthrough point, the study proposed the inducement-type depression and its corresponding typical situations, cognitive and behavioral characteristics of depression, including Academic Achievement-oriented Depression, Physical and Mental Burnout Depression, Peer Relationship-oriented Depression, Family Relationship-oriented Depression, Internet Trauma-oriented Depression, and Physical and Mental Suggestive Depression for students aged 12-24, along with their corresponding typical situations, cognitive, and behavioral characteristics.
Study2 is based on the theory and practical methods of situational tests and aims to develop a depression situational test with ecological validity. After researcher evaluation and multiple revisions using ChatGPT (GPT-3.5 model), 38 initial versions of the depression situational test for students aged 12-24 were generated. Effective data from 160 participants were used for pre-testing of the initial version of the situational test. Eight senior experts from the fields of situational testing, psychological counseling, and clinical assessment evaluated the rationality, typicality, and relevance of the initial test, which included 684 indicators, and confirmed its good internal consistency reliability, content validity, and criterion-related validity, but poor construct validity. After revising based on pre-testing and expert evaluation results, a final version of the depression situational test for students with 40 items was generated. Effective data from 1275 participants were included in the data analysis. The results showed that the final version of the test had good internal consistency reliability, criterion-related validity, and construct validity, confirming the feasibility of jointly assessing the depression situations of adolescents and young people from the perspective of ecological systems and cognitive models. This lays a good practical foundation for developing ecologically-oriented psychological health assessment tools.
Study3 focuses on the psychological needs of adolescents and young people, as well as the needs of related professionals. It aims to explore a digital application of the situational test study. Based on artificial intelligence and psychological counseling technology, the psychological situational data from Study 1 and Study 2 were used for fine-tuning the ChatGLM-6B open-source model, and the AI mental health assistant with fine-tuned model was initially developed and named ”Soulmate”. Six graduate students in the fields of psychology and computer science conducted feasibility testing and verified that Soulmate has a good initial training effect, which is helpful for further generating assistants for psychological health and clinical research, counselor training, and individual psychological support. 
The important significance of this series of research results is to provide a reliable theoretical and practical foundation for building an ecological, intelligent, and personalized psychological health service path. We hope to call on more researchers and practitioners in relevant fields to start from the national conditions, deeply understand the daily psychological needs, and through the triple driving forces of theoretical innovation, technological integration, and practical application, jointly improve the social psychology service system and comprehensively enhance the mental health literacy and quality of life of the public.
 

参考文献总数:

 73    

馆藏号:

 硕045400/23078    

开放日期:

 2024-06-24    

无标题文档

   建议浏览器: 谷歌 360请用极速模式,双核浏览器请用极速模式