Journal article · 2026 · Open access

DisMed-LLM: De-Identifying Spanish Medical Text with Large Language Models

Jesús Alejandro Alzate-Grisales, Joshua Bernal-Salcedo, Jose Manuel Saborit-Torres, Joaquim Montell Serrano, Mariola Penadés Fons, Francisco Garcia‐García, Reinel Tabares-Soto, María de la Iglesia-Vayá

Journal of Healthcare Informatics Research · Published 7 October 2026 · DOI 10.1007/s41666-026-00256-6

Open PDFView at the publisherCite this paper

In brief

“DisMed-LLM: De-Identifying Spanish Medical Text with Large Language Models” is a 2026 journal article by Jesús Alejandro Alzate-Grisales and 7 co-authors, published in Journal of Healthcare Informatics Research. OpenAlex counts 3 citations as of 19 November 2026. It is open access under a CC BY licence: the full text is free to read at Journal of Healthcare Informatics Research.

Citations
3
Most cited in
2026 (2)
Field-weighted impact
2.16
References
27
Authors
8

Abstract

Patient privacy and compliance with regulations such as the General Data Protection Regulation (GDPR) necessitate the robust de-identification of clinical texts. This study investigates the feasibility, computational cost, and limitations of de-identifying Spanish medical records using training-free Large Language Models (LLMs) compared to a supervised baseline. We evaluated five instruction-tuned LLMs on two Spanish clinical corpora (DisMed and MEDDOCAN) and benchmarked them against a fine-tuned multilingual BERT (110M parameters). On a held-out test set, the supervised baseline achieved a strict F1 of 0.904, outperforming the best training-free LLM (Gemma-2 9B, 0.796) while operating roughly 800 times faster. The LLMs’ performance gap was driven entirely by recall and concentrated in categories defined by corpus-specific conventions (e.g., institutions and locations) rather than universal surface forms (names and dates). Testing a newer model (Qwen3-8B) did not resolve this gap, indicating the limitation stems from annotation guidelines rather than model scale. However, when target-corpus annotations were unavailable, the dynamic reversed: in cross-corpus transfer, the supervised model’s F1 collapsed to 0.490, falling below all prompted LLMs, because it rigidly applied the source corpus’s conventions. Finally, novel disclosure metrics revealed that while LLM boundary errors are marginal (0.67%), they miss 26.7% of protected mentions entirely. We conclude that training-free LLMs are a highly adaptable, viable first-pass instrument for institutions lacking annotated data, but their residual disclosure rates mandate human review prior to data release.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Alzate-Grisales, J. A., Bernal-Salcedo, J., Saborit-Torres, J. M., Serrano, J. M., Fons, M. P., Garcia‐García, F., Tabares-Soto, R., & de la Iglesia-Vayá, M. (2026). DisMed-LLM: De-Identifying Spanish Medical Text with Large Language Models. Journal of Healthcare Informatics Research. https://doi.org/10.1007/s41666-026-00256-6

MLA 9

Alzate-Grisales, Jesús Alejandro, et al. “DisMed-LLM: De-Identifying Spanish Medical Text with Large Language Models.” Journal of Healthcare Informatics Research, 2026. https://doi.org/10.1007/s41666-026-00256-6.

Chicago

Alzate-Grisales, Jesús Alejandro, Joshua Bernal-Salcedo, Jose Manuel Saborit-Torres, Joaquim Montell Serrano, Mariola Penadés Fons, Francisco Garcia‐García, Reinel Tabares-Soto, and María de la Iglesia-Vayá. 2026. “DisMed-LLM: De-Identifying Spanish Medical Text with Large Language Models.” Journal of Healthcare Informatics Research. https://doi.org/10.1007/s41666-026-00256-6.

Harvard

Alzate-Grisales, J. A. et al. (2026) ‘DisMed-LLM: De-Identifying Spanish Medical Text with Large Language Models’, Journal of Healthcare Informatics Research. doi:10.1007/s41666-026-00256-6.

IEEE

J. A. Alzate-Grisales et al., “DisMed-LLM: De-Identifying Spanish Medical Text with Large Language Models,” Journal of Healthcare Informatics Research, 2026, doi: 10.1007/s41666-026-00256-6.

BibTeX

@article{alzategrisales2026dismed,
  title     = {{DisMed-LLM: De-Identifying Spanish Medical Text with Large Language Models}},
  author    = {Alzate-Grisales, Jesús Alejandro and Bernal-Salcedo, Joshua and Saborit-Torres, Jose Manuel and Serrano, Joaquim Montell and Fons, Mariola Penadés and Garcia‐García, Francisco and Tabares-Soto, Reinel and de la Iglesia-Vayá, María},
  journal   = {Journal of Healthcare Informatics Research},
  year      = {2026},
  doi       = {10.1007/s41666-026-00256-6},
  url       = {https://doi.org/10.1007/s41666-026-00256-6}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Topic and keywords

Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data

  • de-identification
  • Spanish clinical text
  • large language models
  • patient privacy
  • GDPR compliance
  • medical records
  • annotation guidelines
  • multilingual BERT

Work with this paper

Most-cited papers in Computer Science

All highly cited Computer Science papers

Questions about this paper

How do I cite this paper in APA style?

Alzate-Grisales, J. A., Bernal-Salcedo, J., Saborit-Torres, J. M., Serrano, J. M., Fons, M. P., Garcia‐García, F., Tabares-Soto, R., & de la Iglesia-Vayá, M. (2026). DisMed-LLM: De-Identifying Spanish Medical Text with Large Language Models. Journal of Healthcare Informatics Research. https://doi.org/10.1007/s41666-026-00256-6

How many times has this paper been cited?

3 times, according to OpenAlex on 19 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.

Is this paper free to read?

Yes. It is open access at Journal of Healthcare Informatics Research: sign in and open the PDF from this page.

Who wrote this paper?

Jesús Alejandro Alzate-Grisales, Joshua Bernal-Salcedo, Jose Manuel Saborit-Torres, Joaquim Montell Serrano, Mariola Penadés Fons, Francisco Garcia‐García, Reinel Tabares-Soto, María de la Iglesia-Vayá.

What is the DOI of this paper?

10.1007/s41666-026-00256-6, permanently at https://doi.org/10.1007/s41666-026-00256-6. Sign in and open it from this page.

Bibliographic data from OpenAlex (CC0), last checked 19 November 2026. ResearchGuru does not host this paper; once you are signed in, the links open it at the publisher or the repository that does.