LLM-derived narrative metrics in chronic pain: convergent validity and novel cognitive constructs
The use of subjective reports and questionnaires to evaluate chronic pain, while the gold standard, conveys a limited scope of the patients’ experience and is burdensome. Here we demonstrate that large language models (LLMs) can evaluate patient-provided texts and automatically generate specific, clinically meaningful, and reliable scores for the sensory, affective, and social dimensions of chronic pain. LLMs can also generate assessor-dependent constructs such as agency and personal narrative.
Authors
- Judith Naddour
- Paul Geha (ORCID: https://orcid.org/0000-0002-0537-7216)
- Chadi G. Abdallah (ORCID: https://orcid.org/0000-0001-5783-6181)
- Jennifer S. Gewandter (ORCID: https://orcid.org/0000-0001-7938-6775)
- Guillermo Cecchi (ORCID: https://orcid.org/0000-0003-1013-8348)
- Raquel Norel (ORCID: https://orcid.org/0000-0001-7737-4172)
- Zhiyao Duan (ORCID: https://orcid.org/0000-0002-8334-9974)
- Zhengwu Zhang
Institutions
- University of North Carolina at Chapel Hill (US)
- IBM (United States) (US)
- Baylor College of Medicine (US)
- IBM Research - Thomas J. Watson Research Center (US)
- Applied Physical Sciences (United States) (US)
- University of Rochester (US)
- Icahn School of Medicine at Mount Sinai (US)
Publication Details
- Journal
- npj Digital Medicine
- Published
- 2026-10-06
- DOI
- https://doi.org/10.1038/s41746-026-03359-x
- Primary Topic
- Mental Health via Writing
- Type
- article
- Field-Weighted Citation Impact
- 0.00