Reproduce “Trustworthy AI Psychotherapy: Multi-Agent LLM Workflow for Counseling and Explainable Mental Disorder Diagnosis” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Mithat Can Ozgun, Jiahuan Pei, Koen Hindriks, Lucia Donatelli, Qingzhi Liu, Junxiao Wang
Researchers at Vrije Universiteit Amsterdam introduced DSM5AgentFlow, a multi-agent Large Language Model workflow that autonomously administers DSM-5 Level-1 diagnostic questionnaires. This system generates transparent and explainable mental disorder diagnoses by explicitly linking client conversational evidence to specific DSM-5 criteria, while also simulating realistic therapist-client dialogues for synthetic data generation.
Key insights
- A trade-off exists between LLM conversational fluency and diagnostic reasoning capabilities; models optimized for reasoning achieved higher diagnostic accuracy and explainability, even if conversational quality was slightly lower.
- Embedding intrinsic explainability by explicitly linking AI reasoning to standardized clinical criteria, such as DSM-5, is crucial for fostering trust, clinical adoption, and auditability in sensitive healthcare applications.
- Multi-agent LLM systems can effectively generate large-scale, diversified, and privacy-preserving synthetic therapist-client dialogues, offering a viable solution to the persistent problem of data scarcity in mental health research.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.