International Open Access Journal Platform

青少年心理生态数字孪生体构建及智能预警干预机制创新研究

Innovative Research on the Construction of Digital Twins for Adolescent Psychological Ecosystems and Intelligent Early Warning Intervention Mechanisms

刘曦霞, 陈冠华, 姜安琦, 强景

Psychology of China / 2026,8(6): 941-947 / 2026-06-26 12 8

View Download PDF

• Information:

  济宁医学院，济宁

• Keywords:
  Adolescents; Digital twin; Psychological crisis; Intelligent early warning; Metaverse intervention; Multimodal data fusion
  青少年; 数字孪生; 心理危机; 智能预警; 元宇宙干预; 多模态数据融合
• Abstract: Objective: To address the persistently high detection rate of adolescent depression and the inherent limitations of traditional psychological crisis intervention—characterized by delayed detection and passive response—this study constructs a digital twin-based adolescent psychological ecosystem to establish precision early-warning models and efficient intervention mechanisms. Methods: A multicenter mixed-methods design was employed, recruiting adolescents from three middle schools in Jining, Shandong Province, and randomly assigning them into modeling and validation groups via stratified sampling. Multimodal data—including wearable physiological indicators, campus behavioral records, and social media linguistic features—were integrated to build digital twins using Python-based deep learning frameworks. An improved Long Short-Term Memory (LSTM) network was developed for intelligent early warning, and a VR-based metaverse cognitive behavioral therapy (CBT) intervention pod was constructed. Model performance was evaluated using t-tests, chi-square tests (χ2), multivariable logistic regression, and receiver operating characteristic (ROC) curves. Results: The digital twins demonstrated strong demographic matching and accurate psychological state simulation, with core indicator errors meeting clinical requirements. The improved LSTM model achieved significantly higher AUC than traditional models in the validation group (χ2=42.68, p<0.001). Post-intervention, SDS and SAS scores decreased significantly, with intervention efficacy markedly superior to conventional CBT (t=18.75, p<0.001). Sleep quality, social support, and academic pressure were identified as primary risk factors for psychological crisis (all p<0.001). Conclusion: The integration of digital twin technology with multimodal data enables dynamic, precise simulation and early warning of adolescent psychological states, while the metaverse intervention pod offers an immersive pathway for crisis intervention. The “monitoring-early warning-intervention” full-chain system effectively overcomes traditional limitations, holding significant clinical application value and policy translation potential. 目的：针对青少年抑郁检出率居高不下、传统心理危机干预存在“发现滞后、干预被动”的困境，构建基于数字孪生技术的青少年心理生态系统，建立精准化预警与高效化干预机制。方法：采用多中心混合研究设计，选取济宁市3所中学青少年，分层随机抽样分为建模组与验证组。整合可穿戴设备生理指标、校园行为记录、社交媒体语言特征等多模态数据，运用Python深度学习框架构建数字孪生体，基于改进型LSTM开发智能预警系统，采用VR技术搭建元宇宙认知行为治疗干预舱。运用t检验、卡方检验、logistic回归及ROC曲线评估模型效能。结果：数字孪生体在特征匹配度与状态模拟准确度上表现良好。改进型LSTM预警模型AUC显著优于传统模型（χ2=42.68，p<0.001）。元宇宙干预后SDS、SAS评分显著降低，干预有效率较传统CBT显著提升（t=18.75，p<0.001）。睡眠质量、社交支持、学业压力是心理危机主要影响因素（均p<0.001）。结论：数字孪生技术实现了青少年心理状态动态精准模拟与提前预警，元宇宙干预舱提供了沉浸式干预新路径。“监测—预警—干预”全链条体系有效突破传统模式局限，具有临床应用与政策转化潜力。
• DOI: 10.35534/pc.0806139 (registering DOI)
• Cite: 刘曦霞, 陈冠华, 姜安琦, 强景. (2026). 青少年心理生态数字孪生体构建及智能预警干预机制创新研究. 中国心理学前沿, 8 (6), 941-947.