基于家庭智能语音设备的儿童心理健康预测与干预研究
文献类型：学位论文
| 作者 | 张永艳 |
| 答辩日期 | 2024-06 |
| 文献子类 | 继续教育硕士 |
| 授予单位 | 中国科学院大学 |
| 授予地点 | 中国科学院心理研究所 |
| 其他责任者 | 朱廷劭 |
| 关键词 | 儿童心理健康预测 儿童心理健康干预 家庭智能语音设备 机器学习 聊天机器人 |
| 学位名称 | 理学硕士 |
| 学位专业 | 发展与教育心理学 |
| 其他题名 | A Study of Children's Mental Health Prediction and Intervention Based on Home Intelligent Voice Devices |
| 中文摘要 | Mental health means that all aspects of the mind and the process of activity are in a good or normal state. Mental health is an integral part of health, and children's mental health especially affects every family's heart, especially in the post epidemic era when the seriousness of children's mental health problems is increasing day by day, children's mental health is even more concerned by the society. However, although people's attention to children's mental health has been increasing, and network technology and mobile application technology are also developing rapidly, and various mobile applications are emerging, there are only a limited number of applications focusing on children's mental health, in which parents are actively involved and can be initiated at any time to predict and analyze children's mental health problems. To fill this gap, this paper attempts to explore the construction of a model for predicting and analyzing children's mental health based on the interaction and communication records between children and family smart voice devices, and to design and implement a chatbot to provide interventions for children's mental health problems based on the model's prediction of children's mental health status. This will enhance the initiative of families to participate in the analysis of children's mental health status and make up for the shortcomings of the traditional analysis of children's mental health problems; at the same time, it can also provide a good guarantee of privacy for the analysis of children's mental health. Based on the relevant research theories of children's mental health and the key factors of children's mental health development, this study establishes a metric model of children's mental health, and constructs a lexicon that analyzes children's mental health through the results of interactions with home smart voice devices. Based on the created lexicon, use natural language processing, machine learning, data mining and analysis techniques to design a model for analyzing children's mental health and demonstrate the feasibility of the model. And design relevant experiments and validation schemes to select the optimal scheme among three ways of public, customized and combined dictionaries, as well as four machine learning algorithms, namely, Random Forest, Neural Network Algorithm, Support Vector Machine Regression (SVR) and XGBRegressor. Finally, the chatbot is constructed and trained through artificial intelligence technology so that it can start a dialog with parents, answer their questions about children's mental health problems, and give professional interventions and suggestions. The results of the study show that the difference between the mean and median of the correlation coefficients of the four algorithms of Random Forest, Neural Network, Support Vector Machine Regression (SVR) and XGBRegressor, in terms of the prediction performance statistics of the combination of lexicon, are distributed between 0.04-0.07, the algorithms have stable performance and a balanced distribution, and there are fewer extreme cases. Among the four algorithms, the difference between the mean and the median of the random forest algorithm is significantly lower than the other three algorithms, which has higher stability and accuracy. The results of the intervention study showed that parents recognized the convenience of using chatbots to provide children's mental health interventions as an intervention technique; the chatbots' accuracy in identifying children's mental health problems raised by parents was also relatively high; in terms of the recognition of the effectiveness of the interventions, given that the effectiveness of the interventions needs to be observed over time to determine the effectiveness of the interventions after they are implemented, 32% of the respondents thought that the effectiveness needed to be further observed, and 68% of those who had already implemented it and recognized it as effective temporarily. This paper provides new ideas for the timely detection, ongoing detection, monitoring, and timely intervention of children's mental health problems. The study incorporates an intelligent voice device in the home, aiming to be able to integrate children's mental health assessment into children's daily lives, and to collect the data required for the measurement and complete the assessment without the children's awareness of the intrusion. The innovative technology of providing parents with mental health interventions through chatbots makes it easier for parents to participate in the analysis of children's mental health status, which is no longer limited by the limited resources of psychologists, schools, and hospitals, and provides a small contribution to the early detection of and intervention in children's mental health problems. In addition, the way of analyzing and measuring children's mental health problems in combination with natural language processing, data mining, and mobile applications provides a theoretical space and reference basis for subsequent research in psychology and mobile applications. |
| 英文摘要 | 心理健康是指心理的各个方面及活动过程处于一种良好或正常的状态。心理健康是健康的重要组成部分，儿童的心理健康格外牵动着每一个家庭的心，尤其在儿童心理健康问题的严重性与日俱增的后疫情时代，儿童心理健康更是受到社会的关注。然而虽然人们对于儿童心理健康的关注度不断提升，同时网络技术、移动应用技术也在迅速发展，各种移动应用层出不穷，但专注于儿童心理健康层面，由家长主动参与且可以随时发起预测、分析儿童心理健康问题的应用却十分有限。为填补这一空白，本文将尝试探索基于家庭智能语音设备，构建根据儿童与家庭智能语音设备的互动交流记录来进行儿童心理健康预测分析的模型，并在 模型预测儿童心理健康状态的基础上，设计实现聊天机器人来为儿童心理健康问题提供干预措施。由此提升家庭参与儿童心理健康状态分析的主动性，弥补传统儿童心理健康问题分析中的不足；同时，也能够为儿童心理健康分析提供良好的 私密性保障。 本研究基于儿童心理健康的相关研究理论与儿童心理健康发展的关键因素，建立儿童心理健康的度量模型，构建通过与家庭智能语音设备的交互结果分析儿童心理健康的词典。基于创建的词典，使用自然语言处理、机器学习、数据挖掘和分析等技术，设计儿童心理健康分析模型，论证模型的可行性。并设计相关的实验和验证方案，在公共词典、自定义词典、组合词典三种方式，以及随机森林、神经网络算法、支持向量机回归（SVR）和 XGBRegressor4 种机器学习算法中选 出最优方案。最后，通过人工智能技术，构建和训练聊天机器人，使其可以与家长展开对话，解答家长关于儿童心理健康问题的疑问，给予专业的干预措施和建议。 显示，组合词典方式，在随机森林、神经网络、支持向量机回归（SVR） 和 XGBRegressor4 种算法模型的预测性能统计上，相关系数均值与中值的差，均分布在 0.04-0.07 之间，算法性能稳定且分布比较均衡，极端情况较少。而在 4 种算法之中，随机森林算法的均值与中值差明显低于其它 3 种算法，具有更高的稳定性和准确性。干预方面的研究结果显示，家长对使用聊天机器人提供儿童心理健康干预措施这种干预技术的便利性的认可度非常高；聊天机器人对家长提出的儿童心理健康问题的识别准确性也比较高；在干预措施有效性的认可方面，鉴于干预措施实施后，有效性需要经过长期观察才能确定，所以有 32%的人认为有效性需待进一步观察，已经实施且认可有效的暂占 68%。 本文为及时发现、持续检测、监测以及及时干预儿童心理健康问题提供了新的思路。研究结合了家庭智能语音设备，旨在能够将儿童心理健康测评融入到儿童的日常生活中，在儿童不知不觉中无“侵入”式地完成度量所需数据的收集工作并完成测评。研究创新地通过聊天机器人为家长提供儿童心理健康干预措施的技术，使得儿童心理健康状态分析不再受限于心理医生、学校、医院等有限资源，而是让家长更加便捷、主动地参与进来，为儿童心理健康问题及早发现、及早干预提供了绵薄之力。此外，儿童心理健康问题的分析、度量与自然语言处理、数据挖掘、移动类应用几个层面结合研究的方式，为心理类、移动应用类的后续研 究，提供了理论空间和参考基础。 |
| 语种 | 中文 |
| 源URL | [http://ir.psych.ac.cn/handle/311026/48297]  |
| 专题 | 心理研究所_社会与工程心理学研究室 |
| 推荐引用方式 GB/T 7714 | 张永艳. 基于家庭智能语音设备的儿童心理健康预测与干预研究[D]. 中国科学院心理研究所. 中国科学院大学. 2024. |
入库方式： OAI收割
来源：心理研究所
浏览0
下载0
收藏0
其他版本
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。