arXiv cs.CL· He Hu, Juzheng Si, Qianning Wang, Fei Ma, Fuji Ren, Sicheng Zhao, Yucheng Zhou, Laizhong Cui·· 12 小时前AI 评分31
PsyLLM：融合诊断与治疗推理的心理健康咨询大语言模型
Beyond Empathy: Integrating Diagnostic and Therapeutic Reasoning with Large Language Models for Mental Health Counseling
AI 导读
研究者提出 PsyLLM，一个融合诊断与治疗推理的大语言模型，用于心理健康咨询。该模型基于 Reddit 真实心理困扰帖子构建自动化数据合成管线，依据 DSM/ICD 诊断标准与 CBT、ACT、心理动力学等治疗框架生成结构化临床推理轨迹，并配套新基准与四维评估协议。实验与专业咨询师评估显示，PsyLLM 在该基准上持续优于多个基线模型，代码、数据与模型均已公开。
正文 · 原文
Abstract:Large language models (LLMs) hold significant potential for psychological support, capable of generating empathetic responses and simulating therapeutic conversations. However, existing LLM-based approaches often lack sufficient grounding in structured diagnostic reasoning and diverse therapeutic strategies, limiting their effectiveness in complex counseling scenarios. To address these limitations, we propose PsyLLM, the first large language model designed to incorporate diagnostic- and therapeutic-oriented reasoning for mental health counseling. To develop PsyLLM, we design a novel automated data synthesis pipeline that processes real-world mental health posts collected from Reddit, where users frequently express psychological distress and seek community support. The pipeline transforms these posts into multi-turn counseling dialogues and leverages LLMs guided by international diagnostic standards (e.g., DSM/ICD) and multiple therapeutic frameworks (e.g., CBT, ACT, psychodynamic) to generate structured, clinically informed reasoning traces. A rigorous multi-dimensional filtering mechanism further improves the quality and clinical fidelity of the synthesized dialogue data. In addition, we introduce a new benchmark and evaluation protocol, assessing counseling performance across four key dimensions. Our experiments, along with evaluations by professional counseling experts, demonstrate that PsyLLM consistently outperforms various baseline models on this benchmark. Code, data, and models are available at this https URL.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2505.15715 [cs.CL] |
| (or arXiv:2505.15715v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2505.15715 arXiv-issued DOI via DataCite |
Submission history
From: He Hu [view email]
[v1]
Wed, 21 May 2025 16:24:49 UTC (4,681 KB)
[v2]
Mon, 3 Nov 2025 17:03:30 UTC (3,047 KB)
[v3]
Thu, 5 Nov 2026 16:28:59 UTC (1,745 KB)
来源：arXiv cs.CL · arxiv.org