Training Therapeutic Judges and Multi-Agent Systems for Human-Aligned Mental Health Support
Quick Answer
This paper shows that The TheraJudge and TheraAgent framework enhances mental health support by aligning therapeutic responses with human evaluations, achieving an ICC of 0.87-0.95 with clinicians.
Quick Take
The TheraJudge and TheraAgent framework enhances mental health support by aligning therapeutic responses with human evaluations, achieving an ICC of 0.87-0.95 with clinicians. TheraAgent improves therapeutic quality by +0.43 on a 5-point scale, particularly correcting low-quality responses by +2.45 points, demonstrating the efficacy of human-aligned evaluation in large language models.
Key Points
- TheraJudge is an open-source evaluator trained on human-annotated data across 7 psychological dimensions.
- TheraAgent coordinates responses through roles like Critic, Coach, and Therapist for targeted revisions.
- TheraJudge outperforms supervised baselines in safety, relevance, and empathy evaluations.
- Low-quality responses improved significantly, with a 94% recovery rate after targeted corrections.
- The framework's code is available at https://github.com/vis-nlp/TheraAlign.
Paper Resources
Article Content
From source RSS / original summaryarXiv:2606. 30887v1 Announce Type: new Abstract: Large language models show promise for mental health support, yet therapeutic quality improves only when evaluation functions as an actionable control signal rather than a passive metric. We introduce a framework that formulates therapeutic response generation as a decision-refinement problem driven by multi-dimensional, human-aligned evaluation.
In Stage I, we introduce TheraJudge, an open-source therapeutic evaluator trained via preference-based optimization on human-annotated data to produce reliable judgments across 7 psychological dimensions. In Stage II, we introduce TheraAgent, which operationalizes TheraJudge's evaluations through a coordinated refinement process with specialized Critic, Coach, and Therapist roles that translate evaluative signals into targeted response revisions.
Empirically, TheraJudge achieves strong agreement with clinician ratings, with intraclass correlation coefficients (ICC = 0. 87-0. 95), surpassing supervised baselines and strong closed-source judges, particularly on critical dimensions such as Safety, Relevance, and Empathy. Acting on these evaluations, TheraAgent yields a +0. 43 improvement in human-rated therapeutic quality (on a 5-point scale) under blind evaluation, with 96\% clinician inter-rater reliability. Low-quality responses ($\leq 3$) improve by +2.
45 points with a 94\% recovery rate, demonstrating targeted correction of unsafe outputs. Overall, our results indicate that effective alignment of mental-health LLMs stems from acting on human-aligned evaluation, rather than relying solely on stronger generation. We release code at https://github. com/vis-nlp/TheraAlign.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →Quantifying Prior Dominance in Systems
The study introduces the Normalized Context Utilization (NCU) metric to evaluate Retrieval-Augmented Generation (RAG) systems, revealing that Small Language Models (SLMs) outperform larger models in factual extraction. The findings indicate that traditional scaling laws yield diminishing returns, with a commercial API frequently failing against adversarial evidence due to systemic confidence collapse.