CoChat

Evaluating large language models on medical evidence summarization

Open access

Published version of a preprint

The DOI matches its Crossref record; 29 of the 29 references it cites were found in a registry or an index.

Abstract

Recent advances in large language models (LLMs) have demonstrated remarkable successes in zero- and few-shot performance on various downstream tasks, paving the way for applications in high-stakes domains. In this study, we systematically examine the capabilities and limitations of LLMs, specifically GPT-3.5 and ChatGPT, in performing zero-shot medical evidence summarization across six clinical domains. We conduct both automatic and human evaluations, covering several dimensions of summary quality. Our study demonstrates that automatic metrics often do not strongly correlate with the quality of summaries. Furthermore, informed by our human evaluations, we define a terminology of error types for medical evidence summarization. Our findings reveal that LLMs could be susceptible to generating factually inconsistent summaries and making overly convincing or uncertain statements, leading to potential harm due to misinformation. Moreover, we find that models struggle to identify the salient information and are more error-prone when summarizing over longer textual contexts.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41746-023-00896-7

Cites

29 of 29 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Performance of different summarization systems in automatic evaluations. a Reference-based metrics (higher scores indicate better summaries). b Extractiveness metrics.
Fig. 1 Performance of different summarization systems in automatic evaluations. a Reference-based metrics (higher scores indicate better summaries). b Extractiveness metrics.CC BY · PubMed Central
Fig. 2 Performance of different summarization systems in human evaluations. a Coherence, b factual consistency, c comprehensiveness, and d harmfulness. Statistical analysis by Mann–Whitney U test, *p-value ≤ 0.05, **p-value ≤ 0.01, ***p-value ≤ 0.001, ****p-value ≤ 0.0001.
Fig. 2 Performance of different summarization systems in human evaluations. a Coherence, b factual consistency, c comprehensiveness, and d harmfulness. Statistical analysis by Mann–Whitney U test, *p-value ≤ 0.05, **p-value ≤ 0.01, ***p-value ≤ 0.001, ****p-value ≤ 0.0001.CC BY · PubMed Central
Fig. 3 Annotator vote distribution across all clinical domains and models. a The most and least preferred summaries. b The reasons for choosing the most preferred summaries. c The reasons for choosing the least preferred summaries.
Fig. 3 Annotator vote distribution across all clinical domains and models. a The most and least preferred summaries. b The reasons for choosing the most preferred summaries. c The reasons for choosing the least preferred summaries.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

2 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 29 references

Research works printed but not linked: 2

  1. Lin, C.-Y. ROUGE: a package for automatic evaluation of summaries. In Text Summarization Branches Out. 8 74–81, Barcelona, Spain (Association for Computational Linguistics, 2004).
  2. Banerjee, S. & Lavie, A. METEOR: an automatic metric for MT evaluation with improved correlation with human judgments. In Proc. ACL Workshop on Intrinsic and Extrinsic Evaluation Measures for Machine Translation and/or Summarization 65–72, Ann Arbor, Michigan (Association for Computational Linguistics, 2005).

Cited by

See all 490

Metadata

Authors
Tang, L., Sun, Z., Idnay, B., Nestor, J. G., Soroush, A., Elias, P. A., Xu, Z., Ding, Y., Durrett, G., Rousseau, J. F., Weng, C., & Peng, Y.
Affiliations
  • The University of Texas at Austin (US)
  • Weill Cornell Medicine (US)
  • Columbia University (US)
  • Massachusetts General Hospital (US)
Corresponding author
  • Rousseau, Justin F.
  • Weng, Chunhua
  • Peng, Yifan
Venue
npj Digital Medicine
Published
2023-08-24
Volume
6
Issue
1
Pages
158-158
ISSN
2398-6352 (ISSN-L)
OpenAlex
W4386120650
Cited by
423
References
27
Citations by year
2026: 112 · 2025: 167 · 2024: 122 · 2023: 22
FWCI
16.89
Type
journal article
Language
en
Open access
Licence
Locations
  • npj Digital Medicine (publishedVersion, cc-by)
  • medRxiv (acceptedVersion, cc-by-nc-nd)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (submittedVersion, cc-by-nc-nd)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • U.S. Department of Health & Human Services | NIH | U.S. National Library of Medicine (4R00LM013001, 5R01LM009886)
  • National Science Foundation (2145640, 2019844)
Topics
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
  • Health Sciences › Medicine › Health Informatics › Biomedical Text Mining and Ontologies
Keywords
large language models, zero-shot learning, GPT-3.5, ChatGPT, automatic evaluation metrics, human evaluation, error types, misinformation, clinical domains

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Evaluating large language models on medical evidence summarization. https://app.cochat.ai/papers/10.1038/s41746-023-00896-7
The paper
Tang, L., Sun, Z., Idnay, B., Nestor, J. G., Soroush, A., Elias, P. A., Xu, Z., Ding, Y., Durrett, G., Rousseau, J. F., Weng, C., & Peng, Y. (2023). Evaluating large language models on medical evidence summarization. npj Digital Medicine. https://doi.org/10.1038/s41746-023-00896-7

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.