Scalable information extraction from free text electronic health records using large language models
Open-source LLMs are effective and scalable tools for extracting SDoH from unstructured EHRs, surpassing traditional pattern-matching methods. Further refinement and domain-specific training could enhance their utility in clinical research and predictive analytics, improving healthcare outcomes and addressing health disparities.
Authors
- Rishi Desai (ORCID: https://orcid.org/0000-0003-0299-7273)
- Jie Yang (ORCID: https://orcid.org/0000-0001-5696-363X)
- Santiago Romero‐Brufau (ORCID: https://orcid.org/0000-0002-9922-0083)
- Bowen Gu (ORCID: https://orcid.org/0000-0003-0739-0390)
- Vivian Shao
- Valentina Carducci
- Ziqian Liao (ORCID: https://orcid.org/0009-0003-9985-9172)
Institutions
- Brigham and Women's Hospital (US)
- Mayo Clinic (US)
- Harvard University (US)
- WinnMed (US)
Publication Details
- Journal
- BMC Medical Research Methodology
- Published
- 2025-01-27
- DOI
- https://doi.org/10.1186/s12874-025-02470-z
- Citations
- 22
- Primary Topic
- Topic Modeling
- Type
- article
- Field-Weighted Citation Impact
- 54.58