Review on the Application of Large Language Models in the Mental Health Field

The application of large language models (LLMs) in the mental health field has emerged as a core research direction at the intersection of artificial intelligence (AI) and clinical psychology. This review systematically synthesizes the research progress in this domain from 3 dimensions: model characteristics and empirical evidence, clinical application scenarios, and technological development pathways. At the level of model characteristics and empirical evidence, this paper analyzes the core traits of LLMs and summarizes the empirical support for their applicability in psychological symptom diagnosis and mental illness intervention. In terms of clinical applications, it systematically summarizes the practical cases and application effectiveness of LLMs in scenarios such as mental illness diagnosis, psychological state assessment, virtual psychotherapy, and clinical decision support. From the perspective of technological development, it focuses on sorting out the key advances in directions including data construction, model capability enhancement, and specialized evaluation methods tailored to the mental health field. Finally, the review explicitly points out the core challenges currently facing the research field—such as the disconnect between diagnostic outcomes and clinical practice, insufficient depth of therapeutic simulation, scarcity of high-quality annotated data, and the lack of clinical translation and validation of technologies—and presents an outlook on the future development directions for clinical application implementation and technological innovation research.