基于异常点检测的心理健康辅助诊断方法

扫码查看

原文链接

NETL
NSTL
万方数据

中文摘要：采用异常点检测算法研究心理健康辅助诊断任务,提出并设计一种基于异常点检测的心理健康辅助诊断方法,有效识别心理沙盘数据中的异常样本.在构建心理健康辅助诊断模型过程中,分析数据特性,提取与用户心理健康状况高度相关的特征,构建虚拟心理沙盘数据集;使用4 种传统异常点检测算法,识别沙盘数据集中异常样本,设计融合策略,集成不同算法检测结果,提高异常样本检测精准性和效率,辅助人类专家进行精确诊断;对模型预测性能和结果进行详细分析,结合基线模型进行对比评价.试验结果表明,基于异常点检测的心理健康辅助诊断方法在沙具使用相似度、距离度量、聚类性能等3 项指标上获得较好性能.

外文标题：Approach of assisted diagnosis for mental health based on outlier detection

外文摘要：Towarding the assisted diagnosis for mental health research problem,outlier detection was employed to propose and design an outlier detection-based approach for mental health assisted diagnosis,which could effectively identify outliers from large amounts of virtual mental sandbox samples.During constructing the assisted diagnostic model for mental health,the data characteristics were analyzed,and meanwhile the features highly correlated with the user's mental health status were extracted to establish a virtual psychological sandplay dataset.Four traditional outlier detection algorithms were utilized to identify abnormal samples in the sandplay dataset.A fusion strategy was designed,and the detection results of different algorithms were integrated to enhance the accuracy and efficiency of anomaly sample detection,to assist human experts in making precise diagnoses.The performance and result of the proposed approach were analyzed in details through comparing with the baseline models.The experimental results showed that compared with other models,the assisted diagnosis for mental health based on outlier detection approach had better performance in three indexes such as similarity of sand tool usage,distance of mean vector and clustering performance.

外文关键词：

assisted diagnosis for mental healthvirtual mental sandboxmachine learningoutlier detectionmental health

作者：

乔慧妍、段学龙、解驰皓、赵冬慧、马玉玲

展开 >

作者单位：

山东建筑大学计算机科学与技术学院,山东济南 250101

聆心云(山东)智能科技有限公司,山东济南 250013

关键词：

心理健康辅助诊断虚拟心理沙盘机器学习异常点检测心理健康

基金：

国家自然科学基金资助项目国家自然科学基金资助项目山东省自然科学基金资助项目山东省教育教学研究课题资助项目教育部产学合作协同育人项目2023年度教育部人文社会科学研究专项任务资助项目(高校辅导员研究)2023年度济南市市校融合发展战略工程资助项目

项目编号：

6217703162077033ZR2021MF0442021JXY0122021024230452023JDSZ3174JNSX2023064

出版年：

2024

DOI：

10.6040/j.issn.1672-3961.0.2024.057

山东大学学报(工学版)

山东大学

山东大学学报(工学版)

CSTPCD北大核心

影响因子：0.634

ISSN：1672-3961

年,卷(期)：2024.54(4)