Conference paper · 2026 · Open access

MedPriv-Bench: Benchmarking the Privacy-Utility Trade-off of Large Language Models in Medical Open-Ended Question Answering

GUAN Shaowei, Yukun Zhai, Hin Chi Kwok, Jiawei Du, Xinyu Feng, Jing Li, Harry Qin, Vivian Chi Ching Hui

Conference on Empirical Methods in Natural Language Processing (EMNLP), 26512–26532 · Published 1 January 2026 · DOI 10.18653/2026.emnlp-main.1157

Open PDFView at the publisherCite this paper

In brief

“MedPriv-Bench: Benchmarking the Privacy-Utility Trade-off of Large Language Models in Medical Open-Ended Question Answering” is a 2026 conference paper by GUAN Shaowei and 7 co-authors, published in Conference on Empirical Methods in Natural Language Processing (EMNLP). OpenAlex counts 3 citations as of 19 November 2026, which puts it in the top 4% of works published in 2026 by citations. It is open access: the full text is free to read at Conference on Empirical Methods in Natural Language Processing (EMNLP).

Citations
3
Most cited in
2026 (3)
Field-weighted impact
2.66
References
26
Authors
8

Abstract

Recent advances in Retrieval-Augmented Generation enable LLMs to ground outputs in clinical evidence, but connections to external databases create the risk of contextual leakage, where unique combinations of medical details enable patient re-identification without explicit identifiers.Existing healthcare benchmarks emphasize accuracy while overlooking this risk.To fill this gap, we present MedPriv-Bench 1 , the first benchmark for jointly evaluating privacy preservation and clinical utility in medical open-ended question answering.Our framework utilizes a multi-agent, human-in-the-loop pipeline to synthesize sensitive medical contexts and clinically relevant queries that create realistic privacy pressure.We also establish an automated evaluation protocol using a finetuned RoBERTa-NLI model, which achieved an instance-level F 1 -score of 75.3%, sensitivity of 90.7%, and an average inference time of 0.056 s per sample against human annotations.Across nine LLMs and three privacypreserving methods, we observed a pervasive privacy-utility trade-off.Relative to unprotected Med42-v2-8B (utility 3.87/5; leakage 72.8%), supervised fine-tuning improved utility to 4.25 and reduced leakage to 38.9%, whereas local differential privacy reduced leakage to 20.5% but lowered utility to 3.03.These results demonstrate the need for domain-specific benchmarks to validate medical AI systems in privacy-sensitive settings.

Cite this paper

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

APA 7

Shaowei, G., Zhai, Y., Kwok, H. C., Du, J., Feng, X., Li, J., Qin, H., & Hui, V. C. C. (2026). MedPriv-Bench: Benchmarking the Privacy-Utility Trade-off of Large Language Models in Medical Open-Ended Question Answering. Conference on Empirical Methods in Natural Language Processing (EMNLP), 26512–26532. https://doi.org/10.18653/2026.emnlp-main.1157

MLA 9

Shaowei, GUAN, et al. “MedPriv-Bench: Benchmarking the Privacy-Utility Trade-off of Large Language Models in Medical Open-Ended Question Answering.” Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026, pp. 26512–26532. https://doi.org/10.18653/2026.emnlp-main.1157.

Chicago

Shaowei, GUAN, Yukun Zhai, Hin Chi Kwok, Jiawei Du, Xinyu Feng, Jing Li, Harry Qin, and Vivian Chi Ching Hui. 2026. “MedPriv-Bench: Benchmarking the Privacy-Utility Trade-off of Large Language Models in Medical Open-Ended Question Answering.” Conference on Empirical Methods in Natural Language Processing (EMNLP): 26512–26532. https://doi.org/10.18653/2026.emnlp-main.1157.

Harvard

Shaowei, G. et al. (2026) ‘MedPriv-Bench: Benchmarking the Privacy-Utility Trade-off of Large Language Models in Medical Open-Ended Question Answering’, Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 26512–26532. doi:10.18653/2026.emnlp-main.1157.

IEEE

G. Shaowei et al., “MedPriv-Bench: Benchmarking the Privacy-Utility Trade-off of Large Language Models in Medical Open-Ended Question Answering,” Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 26512–26532, 2026, doi: 10.18653/2026.emnlp-main.1157.

BibTeX

@inproceedings{shaowei2026medpriv,
  title     = {{MedPriv-Bench: Benchmarking the Privacy-Utility Trade-off of Large Language Models in Medical Open-Ended Question Answering}},
  author    = {Shaowei, GUAN and Zhai, Yukun and Kwok, Hin Chi and Du, Jiawei and Feng, Xinyu and Li, Jing and Qin, Harry and Hui, Vivian Chi Ching},
  booktitle = {Conference on Empirical Methods in Natural Language Processing (EMNLP)},
  year      = {2026},
  pages     = {26512--26532},
  doi       = {10.18653/2026.emnlp-main.1157},
  url       = {https://doi.org/10.18653/2026.emnlp-main.1157}
}

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

  • privacy-utility trade-off
  • large language models
  • medical question answering
  • differential privacy
  • retrieval-augmented generation
  • benchmarking
  • artificial intelligence in healthcare
  • clinical utility

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?

Shaowei, G., Zhai, Y., Kwok, H. C., Du, J., Feng, X., Li, J., Qin, H., & Hui, V. C. C. (2026). MedPriv-Bench: Benchmarking the Privacy-Utility Trade-off of Large Language Models in Medical Open-Ended Question Answering. Conference on Empirical Methods in Natural Language Processing (EMNLP), 26512–26532. https://doi.org/10.18653/2026.emnlp-main.1157

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 Conference on Empirical Methods in Natural Language Processing (EMNLP): sign in and open the PDF from this page.

Who wrote this paper?

GUAN Shaowei, Yukun Zhai, Hin Chi Kwok, Jiawei Du, Xinyu Feng, Jing Li, Harry Qin, Vivian Chi Ching Hui.

What is the DOI of this paper?

10.18653/2026.emnlp-main.1157, permanently at https://doi.org/10.18653/2026.emnlp-main.1157. 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.