Academic Journal
Practical Guide to Large Language Models for Information Extraction in Behavioral Health Notes: Tutorial.
| Title: | Practical Guide to Large Language Models for Information Extraction in Behavioral Health Notes: Tutorial. |
|---|---|
| Authors: | Saha D; Department of Computer Science, University of California, Irvine, 6210 Donald Bren Hall, Irvine, CA, 92697, United States, 1 9498247427., Edgcomb JB; Center for Community Health, UCLA Mental Health Informatics & Data Science (MINDS) Hub, University of California, Los Angeles, Los Angeles, CA, United States. |
| Source: | JMIR mental health [JMIR Ment Health] 2026 Sep 24; Vol. 13, pp. e97498. Date of Electronic Publication: 2026 Sep 24. |
| Publication Type: | Journal Article |
| Language: | English |
| Journal Info: | Publisher: JMIR Publications Inc Country of Publication: Canada NLM ID: 101658926 Publication Model: Electronic Cited Medium: Internet ISSN: 2368-7959 (Electronic) Linking ISSN: 23687959 NLM ISO Abbreviation: JMIR Ment Health Subsets: MEDLINE |
| Imprint Name(s): | Original Publication: Toronto : JMIR Publications Inc., [2014]- |
| MeSH Terms: | Information Storage and Retrieval*/methods , Large Language Models* , Electronic Health Records* , Natural Language Processing*, Humans |
| Abstract: | Background: Mental health clinical notes contain decision-critical information often absent from structured electronic health record fields. Large language models (LLMs) can extract clinically relevant signals from narrative text; however, variability in output format, limited reproducibility, and inconsistent evaluation remain barriers to clinical deployment. Despite rapid advances in LLM-based information extraction, clear and reproducible guidance for interdisciplinary clinical teams is limited. Objective: This tutorial aims to present a structured workflow for zero-shot information extraction from mental health clinical notes using locally deployed open-source LLMs. It aims to reduce barriers for clinicians and researchers with limited familiarity with natural language processing (NLP) or LLM-based pipelines. Each stage includes key decision points and examples. The workflow is illustrated on two tasks using synthetic notes: (1) detection of self-injurious thoughts and behaviors (SITB) in pediatric emergency department (ED) notes and (2) antipsychotic medication nonadherence detection in outpatient notes, using schema-constrained outputs and standardized evaluation. Methods: We describe a five-stage zero-shot LLM pipeline: (1) infrastructure setup with local deployment via Ollama to prevent protected health information (PHI) transmission; (2) task definition specifying the clinical construct, output format, and evaluation; (3) dataset preparation using synthetic notes; (4) iterative prompt development using a hold-out development set with binary and Likert scale outputs constrained via JSON schemas; and (5) output parsing, normalization, and validation. We generated 300 synthetic notes per task using separate LLMs for generation and evaluation; 200 notes were used for evaluation, and 100 notes (50 positive and 50 negative) were used as a prompt-development set and excluded from final metrics. Evaluation used Large Language Model Meta AI (Llama) 3.2 and Llama 3.3 with deterministic decoding (temperature=0). Performance was assessed using accuracy, precision, recall, and F Results: We demonstrated the pipeline's functionality using 2 example behavioral health detection tasks. Across both examples, the more capable model (Llama 3.3) performed better than the lighter model used earlier in development (Llama 3.2), and we described how the pipeline's evaluation and error-analysis steps work in practice. These examples also illustrated 2 useful design choices: requiring the model to output in a fixed format reduced errors, and using a graded rating scale, rather than a simple yes/no format, allowed the detection threshold to be adjusted based on clinical risk tolerance. These results are meant to show that the pipeline works as intended, not to serve as a benchmark of real-world accuracy. Conclusions: A schema-driven, zero-shot LLM workflow can support reproducible extraction of clinically relevant information from narrative notes. Local deployment enables processing without transmitting PHI to external servers. This tutorial provides a transferable methodology for institutional adaptation and validation prior to clinical use. All prompts, code, and datasets are publicly available via Zenodo (European Organization for Nuclear Research [CERN]). (© Diya Saha, Juliet B Edgcomb. Originally published in JMIR Mental Health (https://mental.jmir.org).) |
| References: | Br J Psychiatry. 2024 Dec;225(6):532-537. (PMID: 39497458) BMC Med Inform Decis Mak. 2025 Jan 13;25(1):20. (PMID: 39806393) NPJ Digit Med. 2024 Sep 28;7(1):258. (PMID: 39333376) J Med Internet Res. 2025 Jan 23;27:e63126. (PMID: 39847414) Biom J. 2005 Aug;47(4):458-72. (PMID: 16161804) Am J Psychiatry. 2007 Jul;164(7):1035-43. (PMID: 17606655) NPJ Precis Oncol. 2025 Sep 17;9(1):313. (PMID: 40962856) J Med Internet Res. 2022 May 4;24(5):e30405. (PMID: 35507393) J Mood Anxiety Disord. 2025 Jan 31;10:100109. (PMID: 40657592) Sci Rep. 2025 Nov 10;15(1):39231. (PMID: 41213985) Sci Rep. 2025 Nov 24;15(1):45239. (PMID: 41286063) J Am Med Inform Assoc. 2020 Mar 1;27(3):491-497. (PMID: 31682262) JMIR Ment Health. 2026 Feb 04;13:e82371. (PMID: 41637484) J Med Internet Res. 2025 May 5;27:e69284. (PMID: 40324177) J Am Med Inform Assoc. 2025 Jan 1;32(1):218-226. (PMID: 39423850) JMIR Ment Health. 2024 Jul 23;11:e57306. (PMID: 39042893) |
| Contributed Indexing: | Keywords: clinical notes; electronic health records; information extraction; large language models; medication adherence; natural language processing; structured outputs; suicide risk; zero-shot learning |
| Entry Date(s): | Date Created: 20260924 Date Completed: 20260924 Latest Revision: 20260927 |
| Update Code: | 20260927 |
| PubMed Central ID: | PMC13614219 |
| DOI: | 10.2196/97498 |
| PMID: | 42784808 |
| Database: | MEDLINE |
| ISSN: | 2368-7959 |
|---|---|
| DOI: | 10.2196/97498 |