Language-centered mental healthcare with large language models: A comprehensive survey across the care continuum

Mental healthcare faces persistent unmet needs worldwide due to high disease burden and constrained service capacity. Many core functions are language-mediated, including screening and triage, assessment, documentation, longitudinal follow-up, and dialogue-based support, where text and dialogue serve as the primary signal and interface for clinical decision making. Recent advances in large language models, especially long-context processing and instruction following, have accelerated research and early deployments in these language-centered settings, raising urgent questions about clinical integration and assurance. This survey focuses on the language-centered portion of mental healthcare and organizes the literature along a care continuum spanning screening, triage, assessment, diagnosis and risk reasoning, and intervention and decision support, linked by iterative follow-up and language-driven feedback. We conduct a comprehensive survey of studies published between 2019 and October 2025. We (i) trace the evolution from scripted and task-specific NLP systems to LLM-driven approaches, (ii) introduce a three-level taxonomy connecting methodological foundations, service-level applications, and assurance requirements, and (iii) synthesize six major research lines to summarize evidence, identify gaps, and outline deployment-oriented directions for clinically grounded, patient-centered systems.