Evaluation of Alignment Between Large Language Models and Expert Clinicians in Suicide Risk Assessment
LLM-based chatbots' responses to queries aligned with experts' judgment about whether to respond to queries at the extremes of suicide risk (very low and very high), but the chatbots showed inconsistency in addressing intermediate-risk queries, underscoring the need to further refine LLMs.
Authors
- Angela Burnett
- Joshua Breslau (ORCID: https://orcid.org/0000-0002-1194-4643)
- Aaron Kofner (ORCID: https://orcid.org/0000-0001-6980-1218)
- Bradley D. Stein (ORCID: https://orcid.org/0000-0003-1544-458X)
- Hao Yu (ORCID: https://orcid.org/0000-0002-2091-5983)
- Olesya N. Baker (ORCID: https://orcid.org/0000-0001-9125-2761)
- Jonathan Cantor (ORCID: https://orcid.org/0000-0003-4468-833X)
- Li Ang Zhang (ORCID: https://orcid.org/0000-0001-9468-2513)
- Ateev Mehrotra (ORCID: https://orcid.org/0000-0003-2223-1582)
- Ryan K. McBain (ORCID: https://orcid.org/0000-0003-0073-0348)
- Fang Zhang (ORCID: https://orcid.org/0000-0002-3707-9272)
Institutions
- RAND Corporation (US)
- Harvard University (US)
- Brown University (US)
- Harvard Pilgrim Health Care (US)
Publication Details
- Journal
- Psychiatric Services
- Published
- 2025-08-26
- DOI
- https://doi.org/10.1176/appi.ps.20250086
- Citations
- 14
- Primary Topic
- Mental Health via Writing
- Type
- article
- Field-Weighted Citation Impact
- 34.14