The DOI matches its Crossref record; 31 of the 31 references it cites were found in a registry or an index.
Benchmarking large language model-based agent systems for clinical decision tasks
Abstract
Agentic artificial intelligence (AI) systems, designed to autonomously reason, plan, and invoke tools, have shown promise in healthcare, yet systematic benchmarking of their real-world performance remains limited. In this study, we evaluate two such systems: the open-source OpenManus, built on Meta’s Llama-4 and extended with medically customized agents; and Manus, a proprietary agent system employing a multistep planner-executor-verifier architecture. Both systems were assessed across three benchmark families: AgentClinic , a stepwise dialog-based diagnostic simulation; MedAgentsBench , a knowledge-intensive medical QA dataset; and Humanity’s Last Exam (HLE), a suite of challenging text-only and multimodal questions. Despite access to advanced tools (e.g., web browsing, code development and execution, and text file editing) agent systems yielded only modest accuracy gains over baseline LLMs, reaching 60.3% and 28.0% in AgentClinic MedQA and MIMIC, 30.3% on MedAgentsBench, and 8.6% on HLE text. Multimodal accuracy remained low (15.5% on multimodal HLE, 29.2% on AgentClinic NEJM), while resource demands increased substantially, with >10× token usage and >2× latency. Although 89.9% of hallucinations were filtered by in-agent safeguards, hallucinations remained prevalent. These findings reveal that current agentic designs offer modest performance benefits at significant computational and workflow cost, underscoring the need for more accurate, efficient, and clinically viable agent systems.
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
Survey of Hallucination in Natural Language Generation
Ji, Z., Lee, N., Frieske, R., Yu, T., Su, D., Xu, Y., Ishii, E., Bang, Y. J., Madotto, A., & Fung, P. (2023) · ACM Computing Surveys · doi:10.1145/3571730 · Cited by 5,025 · cited 1×
“Previous research has suggested that hallucination is an inherent characteristic of languagegenerationbased onLLMs 28 , andagentsystemsmay amplifythese effects through repeated LLM calls.”
Efficient Memory Management for Large Language Model Serving with PagedAttention
Kwon, W., Li, Z., Zhuang, S., Sheng, Y., Zheng, L., Yu, C. H., Gonzalez, J., Zhang, H., & Stoica, I. (2023) · Proceedings of the 29th Symposium on Operating Systems Principles · doi:10.1145/3600006.3613165 · Cited by 1,817
Efficient Memory Management for Large Language Model Serving with PagedAttention
Kwon, W., Li, Z., Zhuang, S., Sheng, Y., Zheng, L., Yu, C. H., Gonzalez, J. E., Zhang, H., & Stoica, I. (2023) · arXiv · doi:10.48550/arxiv.2309.06180 · Cited by 1,802
Toward expert-level medical question answering with large language models
Singhal, K., Tu, T., Gottweis, J., Sayres, R., Wulczyn, E., Amin, M., Hou, L., Clark, K., Pfohl, S. R., Cole-Lewis, H., et al. (2025) · Nature Medicine · doi:10.1038/s41591-024-03423-7 · Cited by 1,734
Self-citationThe Rise and Potential of Large Language Model Based Agents: A Survey
Xi, Z., Chen, W., Guo, X., He, W., Ding, Y., Hong, B., Zhang, M., Wang, J., Jin, S., Zhou, E., et al. (2023) · arXiv · doi:10.48550/arxiv.2309.07864 · Cited by 972
The rise and potential of large language model based agents: a survey
Xi, Z., Chen, W., Guo, X., He, W., Ding, Y., Hong, B., Zhang, M., Wang, J., Jin, S., Zhou, E., et al. (2025) · Science China Information Sciences · doi:10.1007/s11432-024-4222-0 · Cited by 970
Clinical decision support in the era of artificial intelligence
Shortliffe, E. H., & Sepúlveda, M. J. (2018) · JAMA · doi:10.1001/jama.2018.17163 · Cited by 920
Adapted large language models can outperform medical experts in clinical text summarization
Van Veen, D., Van Uden, C., Blankemeier, L., Delbrouck, J.-B., Aali, A., Bluethgen, C., Pareek, A., Polacin, M., Reis, E. P., Seehofnerová, A., et al. (2024) · Nature Medicine · doi:10.1038/s41591-024-02855-5 · Cited by 891
Bedi, S., Liu, Y., Orr-Ewing, L., Dash, D., Koyejo, S., Callahan, A., Fries, J. A., Wornow, M., Swaminathan, A., Lehmann, L. S., et al. (2024) · doi:10.1101/2024.04.15.24305869 · Cited by 738
Self-citationJin, D., Pan, E., Oufattole, N., Weng, W.-H., Fang, H., & Szolovits, P. (2021) · Applied Sciences · doi:10.3390/app11146421 · Cited by 718
Research works printed but not linked: 22
- Elhaddad, M. & Hamam, S. AI-driven clinical decision support systems: an ongoing pursuit of potential. Cureus 16, e57728 (2024).
- Takita, H. et al. A systematic review and meta-analysis of diagnostic performance comparison between generative AI and physicians. NPJ Digit. Med. 8, 175 (2025). · DOI · Paper page
- Tang, X. et al. MedAgentsBench: benchmarking thinking models and agent frameworks for complex medical reasoning. Preprint at arXiv https://doi.org/10.48550/arXiv.2503.07459 (2025). · DOI · Paper page
- Hong, S. et al. MetaGPT: Meta programming for a multi-agent collaborative framework. In Proc. International Conference Learning Representations Vol. 2024, 23247–23275 (2024).
- Robson, M. J., Xu, S., Wang, Z., Chen, Q. & Ciucci, F. Multi-agent-network-based idea generator for zinc-ion battery electrolyte discovery: a case study on zinc tetrafluoroborate hydrate-based deep eutectic electrolytes. Adv. Mater. 37, e2502649 (2025). · DOI
- Su, H. et al. Many heads are better than one: improved scientific idea generation by A LLM-based multi-agent system. In (eds. Che, W., Nabende, J., Shutova, E. & Pilehvar, M. T.) Proc. 63rd Annual Meeting of the Association for Computational Linguistics, Vol. 1: Long Papers, 28201–28240 (Association for Computational Linguistics, 2025). · DOI
- Li, R. et al. CARE-AD: a multi-agent large language model framework for Alzheimer’s disease prediction using longitudinal clinical notes. NPJ Digit. Med. 8, 541 (2025). · DOI
- Shen, M. & Yang, Q. From mind to machine: the rise of Manus AI as a fully autonomous digital agent. Preprint at arXiv https://doi.org/10.48550/arXiv.2505.02024 (2025). · DOI
- Moritz, M., Topol, E. & Rajpurkar, P. Coordinated AI agents for advancing healthcare. Nat. Biomed. Eng. 9, 432–438 (2025). · DOI
- Wang, Q. et al. AgentTaxo: Dissecting and Benchmarking Token Distribution of LLM Multi-Agent Systems. ICLR 2025 Workshop on Foundation Models in the Wild. https://openreview.net/forum?id=0iLbiYYIpC (2025).
- Fernández-Pichel, M., Pichel, J. C. & Losada, D. E. Evaluating search engines and large language models for answering health questions. NPJ Digit. Med. 8, 153 (2025). · DOI · Paper page
- Song, J., Xu, Z., He, M., Feng, J. & Shen, B. Graph retrieval augmented large language models for facial phenotype associated rare genetic disease. NPJ Digit. Med. 8, 543 (2025). · DOI
- Sim, S. Z. Y. & Chen, T. Critique of impure reason: unveiling the reasoning behaviour of medical large language models. Elife 14, e106187 (2025). · DOI
- Nori, H. et al. Sequential diagnosis with language models. Preprint at arXiv https://doi.org/10.48550/arXiv.2506.22405 (2025). · DOI
- Xu, G. et al. A comprehensive survey of AI Agents in Healthcare. Preprint at TechRxiv https://doi.org/10.36227/techrxiv.176240542.22279040/v2 (2025). · doi:10.36227/techrxiv.176240542.22279040/v2
- Johnson, A. et al. MIMIC-IV (version 3.1). PhysioNet https:https://doi.org/10.13026/kpb9-mt58 (2024). · doi:10.13026/kpb9-mt58
- Johnson, A. E. W. et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci. Data 10, 1 (2023). · DOI · Paper page
- Kim, Y., Wu, J., Abdulle, Y. & Wu, H. MedExQA: medical question answering benchmark with multiple explanations. Preprint at arXiv https://doi.org/10.48550/arXiv.2406.06331 (2024). · DOI · Paper page
- Zuo, Y. et al. MedXpertQA: benchmarking expert-level medical reasoning and understanding. Preprint at arXiv https://doi.org/10.48550/arXiv.2501.18362 (2025). · DOI
- Center for AI Safety, Scale AI & HLE Contributors Consortium. A benchmark of expert-level academic questions toassess AI capabilities. Nature 649, 1139–1146 (2026). · doi:10.1038/s41586-025-09962-4 · Paper page
- Liang, X. et al. OpenManus: an open-source framework for building general AI agents. Preprint at https://doi.org/10.5281/zenodo.15186407 (2025). · doi:10.5281/zenodo.15186407
- Zhu, Y. et al. MedAgentBoard: benchmarking multi-agent collaboration with conventional methods for diverse medical tasks. Preprint at arXiv https://doi.org/10.48550/arXiv.2505.12371 (2025). · DOI · Paper page
Metadata
- Authors
- Liu, Y., Carrero, Z. I., Jiang, X., Ferber, D., Wölflein, G., Zhang, L., Jayabalan, S., Lenz, T., Hui, Z., & Kather, J. N.
- Venue
- npj Digital Medicine
- Published
- 2026-02-18
- Volume
- 9
- Issue
- 1
- ISSN
- 2398-6352
- Cited by
- 37
- References
- 31
- Language
- en
- Licence
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor, 37 d)
- Funders
- German Cancer Aid (DECADE, 70115166)
- German Academic Exchange Service (SECAI, 57616814)
- German Federal Joint Committee (TransplantKI, 01VSF21048)
- European Union’s Horizon Europe research and innovation programme (ODELIA, 101057091; GENIAL, 101096312)
- European Research Council (ERC; NADIR, 101114631)
- National Institute for Health and Care Research (NIHR) Leeds Biomedical Research Centre (NIHR203331)
- German Research Foundation DFG (TRR 412/1, 535081457; SFB 1709/1 2025, 533056198)
- Breast Cancer Research Foundation (BELLADONNA, BCRF-25-225)
- German Federal Ministry of Research, Technology and Space BMFTR (PEARL, 01KD2104C; CAMINO, 01EO2101; TRANSFORM LIVER, 031L0312A; TANGERINE, 01KT2302 through ERA-NET Transcan; Come2Data, 16DKZ2044A; DEEP-HCC, 031L0315A; DECIPHER-M, 01KD2420A; NextBIG, 01ZU2402A)
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: Benchmarking large language model-based agent systems for clinical decision tasks. https://app.cochat.ai/papers/10.1038/s41746-026-02443-6
- The paper
- Liu, Y., Carrero, Z. I., Jiang, X., Ferber, D., Wölflein, G., Zhang, L., Jayabalan, S., Lenz, T., Hui, Z., & Kather, J. N. (2026). Benchmarking large language model-based agent systems for clinical decision tasks. npj Digital Medicine. https://doi.org/10.1038/s41746-026-02443-6
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.