Systematic review outlines applications, risks of large language models in mental health care

A systematic review published in the Journal of Industrial Integration and Management examines how large language models (LLMs) are being applied across mental health care, synthesizing findings from interdisciplinary studies that use diverse data sources such as social media posts, electronic medical records, and multimodal inputs.

The review catalogs applications including social media analysis for early detection of depression, clinical conversational agents, therapy support tools, and psychoeducational content generation, while emphasizing the role of prompt engineering for domain adaptation.

Authors highlight advancements in LLM models and annotation strategies aimed at improving interpretability and clinical relevance, and discuss emerging multimodal fusion techniques that integrate text, speech, and sensor data to enhance diagnosis and monitoring.

The paper also addresses ongoing ethical, sociotechnical, and regulatory challenges, arguing for frameworks that ensure safe, equitable, and accountable deployment of LLMs in real-world mental health care settings.