On the limitations of large language models in clinical diagnosis
Open accessThe DOI matches its Crossref record; 13 of the 13 references it cites were found in a registry or an index.
Abstract
Objective: Large Language Models such as GPT-4 previously have been applied to differential diagnostic challenges based on published case reports. Published case reports have a sophisticated narrative style that is not readily available from typical electronic health records (EHR). Furthermore, even if such a narrative were available in EHRs, privacy requirements would preclude sending it outside the hospital firewall. We therefore tested a method for parsing clinical texts to extract ontology terms and programmatically generating prompts that by design are free of protected health information. Materials and Methods: We investigated different methods to prepare prompts from 75 recently published case reports. We transformed the original narratives by extracting structured terms representing phenotypic abnormalities, comorbidities, treatments, and laboratory tests and creating prompts programmatically. Results: Performance of all of these approaches was modest, with the correct diagnosis ranked first in only 5.3-17.6% of cases. The performance of the prompts created from structured data was substantially worse than that of the original narrative texts, even if additional information was added following manual review of term extraction. Moreover, different versions of GPT-4 demonstrated substantially different performance on this task. Discussion: The sensitivity of the performance to the form of the prompt and the instability of results over two GPT-4 versions represent important current limitations to the use of GPT-4 to support diagnosis in real-life clinical settings. Conclusion: Research is needed to identify the best methods for creating prompts from typically available clinical data to support differential diagnostics.
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
BioBERT: a pre-trained biomedical language representation model for biomedical text mining
Lee, J., Yoon, W., Kim, S., Kim, D., Kim, S., So, C. H., & Kang, J. (2020) · Bioinformatics · doi:10.1093/bioinformatics/btz682 · Cited by 7,218
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445922 · Cited by 6,988
Large language models in medicine
Thirunavukarasu, A. J., Ting, D. S. J., Elangovan, K., Gutierrez, L., Tan, T. F., & Ting, D. S. W. (2023) · Nature Medicine · doi:10.1038/s41591-023-02448-8 · Cited by 3,944
Foundation models for generalist medical artificial intelligence
Moor, M., Banerjee, O., Abad, Z. S. H., Krumholz, H. M., Leskovec, J., Topol, E. J., & Rajpurkar, P. (2023) · Nature · doi:10.1038/s41586-023-05881-4 · Cited by 2,028 · cited 1×
“A po solution could consist in coupling LLMs with a formal representation of medical knowledg example using biomedical knowledge graphs.(6) Future research and algorithmic developm needed to determine the optimal approach to leveraging LLMs for clinical diagnosis.”
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
Expansion of the Human Phenotype Ontology (HPO) knowledge base and resources
Köhler, S., Carmody, L., Vasilevsky, N., Jacobsen, J. O. B., Danis, D., Gourdine, J.-P., Gargano, M., Harris, N. L., Matentzoglu, N., McMurry, J. A., et al. (2019) · Nucleic Acids Research · doi:10.1093/nar/gky1105 · Cited by 677 · cited 1×
“16,18,19 Regular expressions were used to identify phrases that marked separate presentations at specific ages or in specific situations, such as “3 days before this admission” and “After 3 weeks of treatment”.”
Self-citationSelf-supervised learning in medicine and healthcare
Krishnan, R., Rajpurkar, P., & Topol, E. J. (2022) · Nature Biomedical Engineering · doi:10.1038/s41551-022-00914-1 · Cited by 605
Accuracy of a Generative Artificial Intelligence Model in a Complex Diagnostic Challenge
Kanjee, Z., Crowe, B., & Rodman, A. (2023) · JAMA · doi:10.1001/jama.2023.8288 · Cited by 449
Rao, A., Pang, M., Kim, J., Kamineni, M., Lie, W., Prasad, A. K., Landman, A., Dreyer, K., & Succi, M. D. (2023) · Journal of Medical Internet Research · doi:10.2196/48659 · Cited by 412
Towards Expert-Level Medical Question Answering with Large Language Models
Singhal, K., Tu, T., Gottweis, J., Sayres, R., Wulczyn, E., Hou, L., Clark, K., Pfohl, S., Cole-Lewis, H., Neal, D., et al. (2023) · arXiv · doi:10.48550/arxiv.2305.09617 · Cited by 338
Cited by
Moulaei, K., Yadegari, A., Baharestani, M., Farzanbakhsh, S., Sabet, B., & Reza Afrash, M. (2024) · International Journal of Medical Informatics · doi:10.1016/j.ijmedinf.2024.105474 · Cited by 244
Takita, H., Kabata, D., Walston, S. L., Tatekawa, H., Saito, K., Tsujimoto, Y., Miki, Y., & Ueda, D. (2025) · npj Digital Medicine · doi:10.1038/s41746-025-01543-z · Cited by 203
Miao, J., Thongprayoon, C., Suppadungsuk, S., Garcia Valencia, O. A., Qureshi, F., & Cheungpasitporn, W. (2023) · Clinics and Practice · doi:10.3390/clinpract14010008 · Cited by 111
“…are capable of providing personalized medical advice [12], disseminating educational materials and improving medical education [13–15], aiding in clinical decision-making processes [16–18], identifying medical emergencies [19], and providing empathetic responses to patient queries [20–22].”
BackgroundChatGPT-4 Omni Performance in USMLE Disciplines and Clinical Skills: Comparative Analysis
Bicknell, B. T., Butler, D., Whalen, S., Ricks, J., Dixon, C. J., Clark, A. B., Spaedy, O., Skelton, A., Edupuganti, N., Dzubinski, L., et al. (2024) · JMIR Medical Education · doi:10.2196/63430 · Cited by 107
“This dependency on AI tools may lead to a decline in critical thinking and problem-solving abilities, particularly in situations where AI support is unavailable [39,40].”
Analyzing evaluation methods for large language models in the medical field: a scoping review
Lee, J., Park, S., Shin, J., & Cho, B. (2024) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-024-02709-7 · Cited by 52
Metadata
- Authors
- Reese, J. T., Danis, D., Caufield, J. H., Groza, T., Casiraghi, E., Valentini, G., Mungall, C. J., & Robinson, P. N.
- Venue
- medRxiv
- Publisher
- openRxiv
- Published
- 2023-07-14
- OpenAlex
- W4384337834
- Cited by
- 46
- References
- 13
- Type
- posted content
- Language
- en
- Open access
- green
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: On the limitations of large language models in clinical diagnosis. https://app.cochat.ai/papers/10.1101/2023.07.13.23292613
- The paper
- Reese, J. T., Danis, D., Caufield, J. H., Groza, T., Casiraghi, E., Valentini, G., Mungall, C. J., & Robinson, P. N. (2023). On the limitations of large language models in clinical diagnosis. medRxiv. https://doi.org/10.1101/2023.07.13.23292613
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.