数智赋能大学生心理健康监测、预警与干预

Digital Intelligence-Empowered Monitoring, Early Warning, and Intervention for College Students’ Mental Health

摘要: 大学生心理健康是高校育人工作的关键一环，但在具体的心理健康教育实践中面临筛查与预警精度不足、心理数据动态化欠缺、心理风险研判滞后、心理服务供给不足等问题。基于此，本研究提出数智赋能大学生心理健康监测、预警与干预的实践路径：依托大语言模型和检索增强生成技术优化心理筛查精度，借助多源数据融合构建学生心理数字画像，形成“智能初筛 + 人工干预 + 长效追踪”的协同育人模式。研究同时确立伦理知情、隐私安全、人工主导、算法审慎、实用适配五大应用原则，在发挥技术优势的前提下守住伦理与数据安全底线。研究可为高校心理健康教育数字化转型、心理风险前置监测、预警与精准干预提供理论参考与实践方案。

Abstract: College students’ mental health constitutes a foundational pillar of the talent cultivation system in higher education. Nevertheless, prevailing practices in mental health education are confronted with pronounced challenges, including limited screening validity, insufficiency of dynamic behavioral data, delayed risk identification, and structural deficiencies in service provision. To address these predicaments, this study proposes a practical pathway for digital intelligence-empowered monitoring, early warning, and intervention in college students’ mental health. Specifically, the framework integrates large language models (LLMs) with retrieval-augmented generation (RAG) techniques to enhance semantic comprehension and screening precision. Concurrently, multi-source heterogeneous data fusion is adopted to construct personalized psychological digital profiles for individual students, thereby forming a closed-loop collaborative service model characterized by “intelligent preliminary screening—human expert reassessment—long-term follow-up”. To ensure both efficacy and regulatory compliance, this study further articulates five guiding principles: informed ethical consent, data privacy and security, human-centered dominance, algorithmic prudence, and pragmatic adaptability. These principles are designed to strike a dynamic equilibrium between technological enablement and humanistic care. The findings offer both a theoretical reference and a practical paradigm for advancing the digital transformation of mental health services in higher education, with particular emphasis on proactive risk detection and precision-oriented intervention.