{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,23]],"date-time":"2026-10-23T19:51:19Z","timestamp":1792785079221,"version":"4.5.0"},"reference-count":29,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T00:00:00Z","timestamp":1791936000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T00:00:00Z","timestamp":1791936000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"name":"Bayern Innovativ","award":["LSM-2403-0006"],"award-info":[{"award-number":["LSM-2403-0006"]}],"id":[{"id":"https:\/\/ror.org\/047ctny49","id-type":"ROR","asserted-by":"crossref"}]},{"name":"German Research Foundation","award":["TRR 387\/1 - 514894665"],"award-info":[{"award-number":["TRR 387\/1 - 514894665"]}],"id":[{"id":"https:\/\/ror.org\/018mejw64","id-type":"ROR","asserted-by":"crossref"}]},{"name":"DFG BA","award":["537477296"],"award-info":[{"award-number":["537477296"]}],"id":[{"id":"https:\/\/ror.org\/00r0jqe29","id-type":"ROR","asserted-by":"crossref"}]},{"name":"Else-Kr\u00f6ner-Fresenius-Foundation","award":["2025_EKES.03"],"award-info":[{"award-number":["2025_EKES.03"]}]},{"name":"Else-Kr\u00f6ner-Fresenius-Foundation","award":["2024_EKES.16"],"award-info":[{"award-number":["2024_EKES.16"]}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["npj Precis. Onc."],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>\n                    Patients with multiple myeloma accumulate hundreds of clinical documents over years of sequential therapy, and every treatment decision depends on synthesis across that history. Whether large language model\u2013based systems can match expert agreement on this task, and where they fail, has not been established. On longitudinal records of 811 patients treated over 25 years (44,962 documents and 1,334,677 laboratory values), an agentic reasoning system guided by an indexed clinical skill library was compared against single-pass retrieval-augmented generation (RAG), iterative RAG, iterative RAG augmented with the same skill library, and full-context input on 469 expert-adjudicated patient-question pairs, with an exploratory cross-dataset evaluation on MIMIC-IV. The skill-guided agent exceeded the ceiling at which iterative RAG (75.7%) and full-context input (75.8%) converged, reaching 80.1% concordance, with its advantage widening on the most complex tasks (+11.2 percentage points) and longest records. Providing the same skill library to iterative RAG matched the agent in aggregate (79.0%) and on shorter records, whereas the agent retained a significant advantage on the longest records (+10.3 percentage points in the top decile,\n                    <jats:italic>p<\/jats:italic>\n                    \u00a0=\u00a00.017) and a numerical, non-significant advantage on the most complex questions (+5.4 percentage points). The aggregate error rate (12.2%) was comparable to expert disagreement (13.6%), but a higher proportion of system errors were clinically significant (57.8 versus 18.8%).\n                  <\/jats:p>","DOI":"10.1038\/s41698-026-01733-2","type":"journal-article","created":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T20:44:46Z","timestamp":1792010686000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Skill-guided agentic reasoning enables longitudinal synthesis of myeloma records at expert-level concordance"],"prefix":"10.1038","volume":"10","author":[{"given":"Johannes","family":"Moll","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jannik","family":"L\u00fcbberstedt","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Christoph","family":"Nuernbergk","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jacob","family":"Stroh","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Luisa","family":"Mertens","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Anna","family":"Purcarea","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Christopher","family":"Zirn","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zeineb","family":"Benchaaben","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Fabian","family":"Drexel","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hartmut","family":"H\u00e4ntze","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Anirudh","family":"Narayanan","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Friedrich","family":"Puttkammer","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Andrei","family":"Zhukov","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jacqueline","family":"Lammert","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sebastian","family":"Ziegelmayer","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Markus","family":"Graf","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Marion","family":"H\u00f6gner","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Marcus","family":"Makowski","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Florian","family":"Bassermann","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lisa C.","family":"Adams","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jiazhen","family":"Pan","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Daniel","family":"Rueckert","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Krischan","family":"Braitsch","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Keno K.","family":"Bressem","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,10,14]]},"reference":[{"key":"1733_CR1","doi-asserted-by":"publisher","first-page":"1086","DOI":"10.1002\/ajh.26590","volume":"97","author":"SV Rajkumar","year":"2022","unstructured":"Rajkumar, S. V. Multiple myeloma: 2022 update on diagnosis, risk stratification, and management. Am. J. Hematol. 97, 1086\u20131107 (2022).","journal-title":"Am. J. Hematol."},{"key":"1733_CR2","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01965-9","volume":"8","author":"H Cui","year":"2025","unstructured":"Cui, H. et al. Timer: Temporal instruction modeling and evaluation for longitudinal clinical records. npj Digit. Med. 8, 577 (2025).","journal-title":"npj Digit. Med."},{"key":"1733_CR3","doi-asserted-by":"publisher","first-page":"1253","DOI":"10.1093\/jnci\/djaf052","volume":"117","author":"AJ Holmgren","year":"2025","unstructured":"Holmgren, A. J., Apathy, N. C., Crews, J. & Shanafelt, T. National trends in oncology specialists\u2019 EHR inbox work, 2019\u20132022. J. Natl. Cancer Inst. 117, 1253\u20131259 (2025).","journal-title":"J. Natl. Cancer Inst."},{"key":"1733_CR4","doi-asserted-by":"crossref","unstructured":"Du, X. et al. Performance and improvement strategies for adapting generative large language models for electronic health record applications: a systematic review. Int. J. Med. Inform. 205, 106091 (2025).","DOI":"10.1016\/j.ijmedinf.2025.106091"},{"key":"1733_CR5","doi-asserted-by":"publisher","first-page":"605","DOI":"10.1093\/jamia\/ocaf008","volume":"32","author":"S Liu","year":"2025","unstructured":"Liu, S., McCoy, A. B. & Wright, A. Improving large language model applications in biomedicine with retrieval-augmented generation: a systematic review, meta-analysis, and clinical development guidelines. J. Am. Med. Inform. Assoc. 32, 605\u2013615 (2025).","journal-title":"J. Am. Med. Inform. Assoc."},{"key":"1733_CR6","doi-asserted-by":"publisher","first-page":"1418","DOI":"10.1038\/s42256-024-00944-1","volume":"6","author":"J Qiu","year":"2024","unstructured":"Qiu, J. et al. LLM-based agentic systems in medicine and healthcare. Nat. Mach. Intell. 6, 1418\u20131420 (2024).","journal-title":"Nat. Mach. Intell."},{"key":"1733_CR7","doi-asserted-by":"crossref","unstructured":"Truhn, D. et al. Artificial intelligence agents in cancer research and oncology. Nat. Rev. Cancer 26, 256\u2013269 (2026).","DOI":"10.1038\/s41568-025-00900-0"},{"key":"1733_CR8","doi-asserted-by":"publisher","first-page":"AIcs2300235","DOI":"10.1056\/AIcs2300235","volume":"1","author":"D Ferber","year":"2024","unstructured":"Ferber, D. et al. Gpt-4 for information retrieval and comparison of medical oncology guidelines. NEJM AI 1, AIcs2300235 (2024).","journal-title":"NEJM AI"},{"key":"1733_CR9","doi-asserted-by":"publisher","first-page":"e2500011","DOI":"10.1200\/CCI-25-00011","volume":"9","author":"K-H Lu","year":"2025","unstructured":"Lu, K.-H. et al. Enhancing oncology-specific question answering with large language models through fine-tuned embeddings with synthetic data. JCO Clin. Cancer Inform. 9, e2500011 (2025).","journal-title":"JCO Clin. Cancer Inform."},{"key":"1733_CR10","doi-asserted-by":"crossref","unstructured":"Liu, X., Zhang, L., Munir, S., Gu, Y. & Wang, L. Verifact: enhancing long-form factuality evaluation with refined fact extraction and reference facts. In Proc. 2025 Conference on Empirical Methods in Natural Language Processing 17919\u201317936 (Association for Computational Linguistics, 2025).","DOI":"10.18653\/v1\/2025.emnlp-main.905"},{"key":"1733_CR11","first-page":"AIdbp2500418","volume":"3","author":"P Chung","year":"2025","unstructured":"Chung, P. et al. Verifying facts in patient care documents generated by large language models using electronic health records. NEJM AI 3, AIdbp2500418 (2025).","journal-title":"NEJM AI"},{"key":"1733_CR12","doi-asserted-by":"crossref","unstructured":"Myers, S. et al. Evaluating retrieval-augmented generation versus long-context input for clinical reasoning over electronic health records. J. Am. Med. Inform. Assoc. ocag139 (2026).","DOI":"10.1093\/jamia\/ocag139"},{"key":"1733_CR13","doi-asserted-by":"publisher","first-page":"442","DOI":"10.1038\/s41586-025-08866-7","volume":"642","author":"T Tu","year":"2025","unstructured":"Tu, T. et al. Towards conversational diagnostic artificial intelligence. Nature 642, 442\u2013450 (2025).","journal-title":"Nature"},{"key":"1733_CR14","unstructured":"Schmidgall, S. et al. Agentclinic: a multimodal agent benchmark to evaluate AI in simulated clinical environments. Preprint at arXiv:2405.07960 (2024)."},{"key":"1733_CR15","doi-asserted-by":"crossref","unstructured":"Shi, W. et al. EHRAgent: code empowers large language models for few-shot complex tabular reasoning on electronic health records. In Proc. 2024 Conference on Empirical Methods in Natural Language Processing 22315\u201322339 (Association for Computational Linguistics, 2024).","DOI":"10.18653\/v1\/2024.emnlp-main.1245"},{"key":"1733_CR16","doi-asserted-by":"publisher","first-page":"AIdbp2500144","DOI":"10.1056\/AIdbp2500144","volume":"2","author":"Y Jiang","year":"2025","unstructured":"Jiang, Y. et al. MedAgentBench: a virtual EHR environment to benchmark medical LLM agents. NEJM AI 2, AIdbp2500144 (2025).","journal-title":"NEJM AI"},{"key":"1733_CR17","unstructured":"Lee, G. et al. FHIR-AgentBench: benchmarking LLM agents for realistic interoperable EHR question answering. In Proc. Fifth Machine Learning for Health Symposium 1297\u20131315 (PMLR, 2026)."},{"key":"1733_CR18","doi-asserted-by":"publisher","unstructured":"Johnson, A., Pollard, T., Horng, S., Celi, L. A. & Mark, R. MIMIC-IV-note: deidentified free-text clinical notes. PhysioNet. https:\/\/doi.org\/10.13026\/1n74-ne17 (2023).","DOI":"10.13026\/1n74-ne17"},{"key":"1733_CR19","doi-asserted-by":"publisher","DOI":"10.1038\/s41597-022-01899-x","volume":"10","author":"AE Johnson","year":"2023","unstructured":"Johnson, A. E. et al. Mimic-iv, a freely accessible electronic health record dataset. Sci. Data 10, 1 (2023).","journal-title":"Sci. Data"},{"key":"1733_CR20","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01475-8","volume":"8","author":"AJ Goodell","year":"2025","unstructured":"Goodell, A. J., Chu, S. N., Rouholiman, D. & Chu, L. F. Large language model agents can use tools to perform clinical calculations. npj Digit. Med. 8, 163 (2025).","journal-title":"npj Digit. Med."},{"key":"1733_CR21","doi-asserted-by":"publisher","first-page":"491","DOI":"10.1038\/s43856-025-01194-x","volume":"5","author":"GA Prenosil","year":"2025","unstructured":"Prenosil, G. A. et al. Neuro-symbolic AI for auditable cognitive information extraction from medical reports. Commun. Med. 5, 491 (2025).","journal-title":"Commun. Med."},{"key":"1733_CR22","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01670-7","volume":"8","author":"E Asgari","year":"2025","unstructured":"Asgari, E. et al. A framework to assess clinical safety and hallucination rates of LLMs for medical text summarisation. NPJ Digit. Med. 8, 274 (2025).","journal-title":"NPJ Digit. Med."},{"key":"1733_CR23","doi-asserted-by":"publisher","first-page":"157","DOI":"10.1162\/tacl_a_00638","volume":"12","author":"NF Liu","year":"2024","unstructured":"Liu, N. F. et al. Lost in the middle: how language models use long contexts. Trans. Assoc. Comput. Linguist. 12, 157\u2013173 (2024).","journal-title":"Trans. Assoc. Comput. Linguist."},{"key":"1733_CR24","unstructured":"Normand, O. et al. A real-world evaluation of LLM medication safety reviews in NHS primary care. Preprint at arXiv:2512.21127 (2025)."},{"key":"1733_CR25","doi-asserted-by":"publisher","first-page":"e215","DOI":"10.1161\/01.CIR.101.23.e215","volume":"101","author":"AL Goldberger","year":"2000","unstructured":"Goldberger, A. L. et al. Physiobank, physiotoolkit, and physionet: components of a new research resource for complex physiologic signals. Circulation 101, e215\u2013e220 (2000).","journal-title":"Circulation"},{"key":"1733_CR26","unstructured":"Agarwal, S. et al. gpt-oss-120b & gpt-oss-20b model card. Preprint at arXiv:2508.10925 (2025)."},{"key":"1733_CR27","first-page":"9459","volume":"33","author":"P Lewis","year":"2020","unstructured":"Lewis, P. et al. Retrieval-augmented generation for knowledge-intensive NLP tasks. Adv. Neural Inf. Process. Syst. 33, 9459\u20139474 (2020).","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"1733_CR28","unstructured":"Doan, N. N., H\u00e4rm\u00e4, A., Celebi, R. & Gottardo, V. A hybrid retrieval approach for advancing retrieval-augmented generation systems. In Proc. 7th International Conference on Natural Language and Speech Processing (ICNLSP 2024) 397\u2013409 (Association for Computational Linguistics, 2024)."},{"key":"1733_CR29","unstructured":"Collins, G. S. et al. Tripod\u2009+\u2009AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ 385, e078378 (2024)."}],"container-title":["npj Precision Oncology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s41698-026-01733-2","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41698-026-01733-2.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41698-026-01733-2.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T20:45:43Z","timestamp":1792010743000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s41698-026-01733-2"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,10,14]]},"references-count":29,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["1733"],"URL":"https:\/\/doi.org\/10.1038\/s41698-026-01733-2","relation":{},"ISSN":["2397-768X"],"issn-type":[{"value":"2397-768X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10,14]]},"assertion":[{"value":"8 June 2026","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"23 September 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"14 October 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"A.P. has received travel support from Janssen-Cilag and Kite\/Gilead, but declares no non-financial competing interests. F.B. has received honoraria from Amgen, Johnson & Johnson, Bristol Myers Squibb, AbbVie, and GSK (personal and institutional) and travel support from Amgen, Johnson & Johnson, and Bristol Myers Squibb (institutional), and has participated on advisory boards for Amgen, Johnson & Johnson, Bristol Myers Squibb, AbbVie, and GSK (personal), but declares no non-financial competing interests. M.H. has received honoraria from Johnson & Johnson, Sanofi, GSK, Oncopeptides, and Pfizer, payment for expert testimony from Johnson & Johnson and Oncopeptides, and travel support from Johnson & Johnson, Oncopeptides, Pfizer, and Amgen, and has participated on advisory boards for Johnson & Johnson, Sanofi, Oncopeptides, and Pfizer, but declares no non-financial competing interests. J.La has received grants or contracts from the Federal Ministry of Research, Technology and Space, the Bavarian State Ministry of Science and the Arts, the Bavarian State Ministry of Health, Care and Prevention, the Federal Joint Committee (G-BA), the Association for Molecular Pathology, and the German Cancer Consortium, a Google Gemma Academic Program Award, and honoraria from the Forum for Continuing Medical Education (Germany), AstraZeneca (Germany), and Novartis (Germany), and has participated on an advisory board for Novartis (Germany), but declares no non-financial competing interests. The remaining authors declare no financial or non-financial competing interests. None of the authors is an Editor-in-Chief, Associate Editor, or Guest Editor of npj Precision Oncology.","order":1,"name":"Ethics","label":"Competing interests","group":{"name":"EthicsHeading","label":"Declarations"}}],"article-number":"389"}}