Wang Xi, Zhou Yujia, Zhou Guangyu
School of Psychological and Cognitive Sciences, Beijing Key Laboratory of Behavior and Mental Health, Key Laboratory of Machine Perception (Ministry of Education), Peking University, Beijing, China.
Department of Computer Science and Technology, Tsinghua University, Beijing, China.
JMIR Ment Health. 2025 Jun 27;12:e70610. doi: 10.2196/70610.
Mental health disorders affect an estimated 1 in 8 individuals globally, yet traditional interventions often face barriers, such as limited accessibility, high costs, and persistent stigma. Recent advancements in generative artificial intelligence (GenAI) have introduced AI systems capable of understanding and producing humanlike language in real time. These developments present new opportunities to enhance mental health care.
We aimed to systematically examine the current applications of GenAI in mental health, focusing on 3 core domains: diagnosis and assessment, therapeutic tools, and clinician support. In addition, we identified and synthesized key ethical issues reported in the literature.
We conducted a comprehensive literature search, following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines, in PubMed, ACM Digital Library, Scopus, Embase, PsycInfo, and Google Scholar databases to identify peer-reviewed studies published from October 1, 2019, to September 30, 2024. After screening 783 records, 79 (10.1%) studies met the inclusion criteria.
The number of studies on GenAI applications in mental health has grown substantially since 2023. Studies on diagnosis and assessment (37/79, 47%) primarily used GenAI models to detect depression and suicidality through text data. Studies on therapeutic applications (20/79, 25%) investigated GenAI-based chatbots and adaptive systems for emotional and behavioral support, reporting promising outcomes but revealing limited real-world deployment and safety assurance. Clinician support studies (24/79, 30%) explored GenAI's role in clinical decision-making, documentation and summarization, therapy support, training and simulation, and psychoeducation. Ethical concerns were consistently reported across the domains. On the basis of these findings, we proposed an integrative ethical framework, GenAI4MH, comprising 4 core dimensions-data privacy and security, information integrity and fairness, user safety, and ethical governance and oversight-to guide the responsible use of GenAI in mental health contexts.
GenAI shows promise in addressing the escalating global demand for mental health services. They may augment traditional approaches by enhancing diagnostic accuracy, offering more accessible support, and reducing clinicians' administrative burden. However, to ensure ethical and effective implementation, comprehensive safeguards-particularly around privacy, algorithmic bias, and responsible user engagement-must be established.
据估计，全球每8个人中就有1人受到心理健康障碍的影响，但传统干预措施往往面临诸多障碍，如可及性有限、成本高昂以及持续存在的污名化问题。生成式人工智能（GenAI）的最新进展引入了能够实时理解和生成类人语言的人工智能系统。这些发展为加强心理健康护理带来了新机遇。
我们旨在系统地研究GenAI在心理健康领域的当前应用，重点关注三个核心领域：诊断与评估、治疗工具以及临床医生支持。此外，我们还识别并综合了文献中报道的关键伦理问题。
我们按照PRISMA（系统评价和Meta分析优先报告项目）2020指南，在PubMed、ACM数字图书馆、Scopus、Embase、PsycInfo和谷歌学术数据库中进行了全面的文献检索，以识别2019年10月1日至2024年9月30日期间发表的同行评审研究。在筛选了783条记录后，79项（10.1%）研究符合纳入标准。
自2023年以来，关于GenAI在心理健康领域应用的研究数量大幅增长。诊断与评估方面的研究（37/79，47%）主要使用GenAI模型通过文本数据检测抑郁症和自杀倾向。治疗应用方面的研究（20/79，25%）调查了基于GenAI的聊天机器人和用于情感与行为支持的自适应系统，报告了有前景的结果，但显示出实际应用有限且缺乏安全保障。临床医生支持方面的研究（24/79，30%）探讨了GenAI在临床决策、文档记录与总结、治疗支持、培训与模拟以及心理教育中的作用。各领域均一致报告了伦理问题。基于这些发现，我们提出了一个综合伦理框架GenAI4MH，它包含四个核心维度——数据隐私与安全、信息完整性与公平性、用户安全以及伦理治理与监督——以指导在心理健康背景下负责任地使用GenAI。
GenAI在满足全球对心理健康服务不断增长的需求方面显示出前景。它们可以通过提高诊断准确性、提供更易获得的支持以及减轻临床医生的行政负担来增强传统方法。然而，为确保伦理且有效的实施，必须建立全面的保障措施，特别是围绕隐私、算法偏差和负责任的用户参与方面。
