Comparative study of ChatGPT and human evaluators on the assessment of medical literature according to recognised reporting standards
Open accessThe DOI matches its Crossref record; 7 of the 7 references it cites were found in a registry or an index.
Abstract
Introduction: Amid clinicians’ challenges in staying updated with medical research, artificial intelligence (AI) tools like the large language model (LLM) ChatGPT could automate appraisal of research quality, saving time and reducing bias. This study compares the proficiency of ChatGPT3 against human evaluation in scoring abstracts to determine its potential as a tool for evidence synthesis. Methods: We compared ChatGPT’s scoring of implant dentistry abstracts with human evaluators using the Consolidated Standards of Reporting Trials for Abstracts reporting standards checklist, yielding an overall compliance score (OCS). Bland-Altman analysis assessed agreement between human and AI-generated OCS percentages. Additional error analysis included mean difference of OCS subscores, Welch’s t-test and Pearson’s correlation coefficient. Results: Bland-Altman analysis showed a mean difference of 4.92% (95% CI 0.62%, 0.37%) in OCS between human evaluation and ChatGPT. Error analysis displayed small mean differences in most domains, with the highest in ‘conclusion’ (0.764 (95% CI 0.186, 0.280)) and the lowest in ‘blinding’ (0.034 (95% CI 0.818, 0.895)). The strongest correlations between were in ‘harms’ (r=0.32, p<0.001) and ‘trial registration’ (r=0.34, p=0.002), whereas the weakest were in ‘intervention’ (r=0.02, p<0.001) and ‘objective’ (r=0.06, p<0.001). Conclusion: LLMs like ChatGPT can help automate appraisal of medical literature, aiding in the identification of accurately reported research. Possible applications of ChatGPT include integration within medical databases for abstract evaluation. Current limitations include the token limit, restricting its usage to abstracts. As AI technology advances, future versions like GPT4 could offer more reliable, comprehensive evaluations, enhancing the identification of high-quality research and potentially improving patient outcomes.
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.
Integrity signals
Registry
Cites
Statistics
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.
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
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929
Benefits, Limits, and Risks of GPT-4 as an AI Chatbot for Medicine
Lee, P., Bubeck, S., & Petro, J. (2023) · New England Journal of Medicine · doi:10.1056/nejmsr2214184 · Cited by 1,738
Performance of GPT-3.5 and GPT-4 on the Japanese Medical Licensing Examination: Comparison Study
Takagi, S., Watari, T., Erabi, A., & Sakaguchi, K. (2023) · JMIR Medical Education · doi:10.2196/48002 · Cited by 271
HS Kumar, A. (2023) · Biology, Engineering, Medicine and Science Reports · doi:10.5530/bems.9.1.5 · Cited by 165
Dr ChatGPT tell me what I want to hear: How different prompts impact health answer correctness
Koopman, B., & Zuccon, G. (2023) · Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2023.emnlp-main.928 · Cited by 73
Reporting quality of abstracts of randomized controlled trials related to implant dentistry
Menne, M. C., Pandis, N., & Faggion, C. M. (2022) · Journal of Periodontology · doi:10.1002/jper.21-0396 · Cited by 8
Cited by
Scherbakov, D., Hubig, N., Jansari, V., Bakumenko, A., & Lenert, L. A. (2024) · arXiv · doi:10.48550/arxiv.2409.04600 · Cited by 156
Scherbakov, D., Hubig, N., Jansari, V., Bakumenko, A., & Lenert, L. A. (2025) · Journal of the American Medical Informatics Association · doi:10.1093/jamia/ocaf063 · Cited by 154
Cao, J. J., Kwon, D. H., Ghaziani, T. T., Kwo, P., Tse, G., Kesselman, A., Kamaya, A., & Tse, J. R. (2024) · Abdominal Radiology · doi:10.1007/s00261-024-04501-7 · Cited by 28
Forero, D. A., Abreu, S. E., Tovar, B. E., & Oermann, M. H. (2025) · Journal of Medical Systems · doi:10.1007/s10916-025-02212-0 · Cited by 13
“For the statistical analysis, several complementary approaches were used [21].”
Metadata
- Authors
- Roberts, R. H., Ali, S. R., Hutchings, H. A., Dobbs, T. D., & Whitaker, I. S.
- Venue
- BMJ Health & Care Informatics
- Publisher
- BMJ
- Published
- 2023-10
- Volume
- 30
- Issue
- 1
- Pages
- e100830
- ISSN
- 2632-1009 (ISSN-L)
- OpenAlex
- W4387599176
- Cited by
- 36
- References
- 7
- FWCI
- 6.61
- Type
- journal article
- Language
- en
- Open access
- gold
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: Comparative study of ChatGPT and human evaluators on the assessment of medical literature according to recognised reporting standards. https://app.cochat.ai/papers/10.1136/bmjhci-2023-100830
- The paper
- Roberts, R. H., Ali, S. R., Hutchings, H. A., Dobbs, T. D., & Whitaker, I. S. (2023). Comparative study of ChatGPT and human evaluators on the assessment of medical literature according to recognised reporting standards. BMJ Health & Care Informatics. https://doi.org/10.1136/bmjhci-2023-100830
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.