{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,20]],"date-time":"2026-08-20T16:30:08Z","timestamp":1787243408321,"version":"build-2736575974"},"reference-count":24,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["JMIR AI"],"abstract":"Abstract<\/jats:title>\n \n Background<\/jats:title>\n Multidisciplinary tumor boards (MDTs) constitute the foundation of modern tumor therapy. Large language models (LLMs) are widely discussed for optimizing their recommendations.<\/jats:p>\n <\/jats:sec>\n \n Objective<\/jats:title>\n This is the first prospective feasibility study evaluating the implementation of locally run LLMs on real-world cases within a regular head and neck MDT.<\/jats:p>\n <\/jats:sec>\n \n Methods<\/jats:title>\n Seventeen patients participated in the study. The MDT cases were processed by 2 different local LLMs (gemma-3-12b and gpt-oss-20b) to obtain treatment recommendations. The MDT conferred as usual. After the decision was made, the MDT was presented with the LLMs\u2019 recommendations. If deemed to be beneficial, the MDT\u2019s recommendation was adjusted. The MDT members rated the LLMs\u2019 responses inter alia, for medical adequacy on a 6-point Likert scale. In addition, a tabular comparison of the MDT\u2019s and LLMs\u2019 recommendations was carried out.<\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n In one case, 6% (1\/17, 95% CI 0%\u201029%), the LLM was able to substantially improve the MDT recommendation by underscoring a follow-up examination that had not yet been performed. Concordance regarding the curative or palliative therapy regimen reached 94% (16\/17, 95% CI 71%\u2010100%); for gemma-3-12b and 59% (10\/17, 95% CI 33%\u201082%) for gpt-oss-20b. Gemma-3-12b stated the same first-line therapy regimen as the MDT as first-line in 35% (6\/17, 95% CI 14%\u201062%) of cases, and gpt-oss-20b in 41% (7\/17, 95% CI 18%\u201067%) of cases. In 59% (10\/17, 95% CI 33%\u201082%) of patients, gemma-3-12b stated the MDT\u2019s first-line therapy regimen, albeit with a different priority, while for gpt-oss-20b, it was 41% (7\/17, 95% CI 18%\u201067%) of patients. Medical adequacy, as rated by the MDT members, revealed a median of 5 (IQR 2\u20105) for gemma-3-12b and 4 (IQR 3\u20105) for gpt-oss-20b. MDT members stated potentially hazardous information in 27% (25\/93, 95% CI 18%\u201037%) of ratings for gemma-3-12b and 17% (14\/83, 95% CI 9%\u201026%) of ratings for gpt-oss-20b.<\/jats:p>\n <\/jats:sec>\n \n Conclusions<\/jats:title>\n Locally run LLMs improved the MDT recommendation in 1 case and were primarily useful for identifying potentially relevant missing information in other cases, underscoring that they cannot replace MDTs. However, their observed benefit suggests that more advanced local models may offer safe, rapid, and cost-effective support for MDT decision-making. The study should be seen as an exploratory setting focusing on practical insights rather than benchmarking its clinical impact. Accordingly, the data demonstrate that the integration of LLMs in today\u2019s MDT workflow is feasible and may benefit the quality of decision-making in specific cases.<\/jats:p>\n <\/jats:sec>","DOI":"10.2196\/95081","type":"journal-article","created":{"date-parts":[[2026,8,20]],"date-time":"2026-08-20T16:22:19Z","timestamp":1787242939000},"page":"e95081-e95081","source":"Crossref","is-referenced-by-count":0,"title":["Augmenting Head and Neck Multidisciplinary Tumor Board Recommendations With Locally Run Large Language Models: Prospective Evaluation of Real-World Implementation"],"prefix":"10.2196","volume":"5","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-9551-2310","authenticated-orcid":false,"given":"Christoph Raphael","family":"Buhr","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8626-4044","authenticated-orcid":false,"given":"Lukas","family":"M\u00fcller","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4785-6394","authenticated-orcid":false,"given":"Daniel","family":"Pinto dos Santos","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4081-1487","authenticated-orcid":false,"given":"Daniel","family":"Thiem","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8950-5723","authenticated-orcid":false,"given":"Jean-Jacques","family":"Ponciano","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Maximilian","family":"Kr\u00fcger","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7444-710X","authenticated-orcid":false,"given":"Karoline","family":"O'Brien","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0189-3499","authenticated-orcid":false,"given":"Justus","family":"Kaufmann","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hildegard","family":"Nolte","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4737-9305","authenticated-orcid":false,"given":"Martin","family":"Gartenschl\u00e4eger","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Stefanie","family":"Zimmer","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sebastian","family":"Altmann","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7913-6872","authenticated-orcid":false,"given":"Andrew","family":"Blaikie","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Christian","family":"Ruckes","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7569-2393","authenticated-orcid":false,"given":"Christoph","family":"Matthias","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8031-2973","authenticated-orcid":false,"given":"Sebastian","family":"Kuhn","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5498-4031","authenticated-orcid":false,"given":"Jonas","family":"Eckrich","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,8,20]]},"reference":[{"issue":"4","key":"R1","doi-asserted-by":"publisher","first-page":"2105","DOI":"10.1007\/s00405-023-08326-w","article-title":"Accuracy of ChatGPT in head and neck oncological board decisions: preliminary findings","volume":"281","author":"Lechien","journal-title":"Eur Arch Otorhinolaryngol"},{"key":"R2","doi-asserted-by":"publisher","DOI":"10.3389\/fonc.2024.1353031","article-title":"Assessing the role of advanced artificial intelligence as a tool in multidisciplinary tumor board decision-making for primary head and neck cancer cases","volume":"14","author":"Schmidl","journal-title":"Front Oncol"},{"issue":"11","key":"R3","doi-asserted-by":"publisher","first-page":"6099","DOI":"10.1007\/s00405-024-08828-1","article-title":"Assessing the use of the novel tool Claude 3 in comparison to ChatGPT 4.0 as an artificial intelligence tool in the diagnosis and therapy of primary head and neck cancer cases","volume":"281","author":"Schmidl","journal-title":"Eur Arch Otorhinolaryngol"},{"issue":"9","key":"R4","doi-asserted-by":"publisher","first-page":"4815","DOI":"10.1007\/s00405-025-09524-4","article-title":"ChatGPT versus DeepSeek in head and neck cancer staging and treatment planning: guideline-based study","volume":"282","author":"Vural Camalan","journal-title":"Eur Arch Otorhinolaryngol"},{"key":"R5","unstructured":"Health Insurance Portability and Accountability Act of 1996. U.S. Government Publishing Office (GPO) \/ U.S. Congress; 1996. URL: https:\/\/www.congress.gov\/104\/plaws\/publ191\/PLAW-104publ191.pdf [Accessed 28-07-2026]"},{"key":"R6","unstructured":"Regulation (EU) 2016\/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing directive 95\/46\/EC (General Data Protection Regulation) (text with EEA relevance). Publications Office of the European Union; 2016. URL: https:\/\/eur-lex.europa.eu\/eli\/reg\/2016\/679\/oj\/eng [Accessed 28-07-2026]"},{"issue":"3","key":"R7","doi-asserted-by":"publisher","first-page":"1593","DOI":"10.1007\/s00405-024-09153-3","article-title":"Assessment of decision-making with locally run and web-based large language models versus human board recommendations in otorhinolaryngology, head and neck surgery","volume":"282","author":"Buhr","journal-title":"Eur Arch Otorhinolaryngol"},{"issue":"3","key":"R8","doi-asserted-by":"publisher","first-page":"1619","DOI":"10.1007\/s00405-024-08947-9","article-title":"Prediction of tumor board procedural recommendations using large language models","volume":"282","author":"Aubreville","journal-title":"Eur Arch Otorhinolaryngol"},{"issue":"9","key":"R9","doi-asserted-by":"publisher","first-page":"3049","DOI":"10.1002\/lary.32198","article-title":"Applications of natural language processing in otolaryngology: a scoping review","volume":"135","author":"Banyi","journal-title":"Laryngoscope"},{"issue":"1","key":"R10","doi-asserted-by":"publisher","DOI":"10.1186\/s12911-025-02954-4","article-title":"A systematic review of large language model (LLM) evaluations in clinical medicine","volume":"25","author":"Shool","journal-title":"BMC Med Inform Decis Mak"},{"key":"R11","doi-asserted-by":"publisher","DOI":"10.2196\/59505","article-title":"Multimodal large language models in health care: applications, challenges, and future outlook","volume":"26","author":"AlSaad","journal-title":"J Med Internet Res"},{"issue":"9","key":"R12","doi-asserted-by":"publisher","first-page":"2613","DOI":"10.1038\/s41591-024-03097-1","article-title":"Evaluation and mitigation of the limitations of large language models in clinical decision-making","volume":"30","author":"Hager","journal-title":"Nat Med"},{"key":"R13","unstructured":"DeepL [Website in German]. URL: https:\/\/www.deepl.com\/de [Accessed 12-05-2025]"},{"key":"R14","unstructured":"Calculate confidence limits for a sample proportion. Epitools. 2026. URL: https:\/\/epitools.ausvet.com.au\/ciproportion [Accessed 12-05-2026]"},{"issue":"9","key":"R15","doi-asserted-by":"publisher","DOI":"10.1007\/s10462-024-10896-y","article-title":"Towards trustworthy LLMs: a review on debiasing and dehallucinating in large language models","volume":"57","author":"Lin","journal-title":"Artif Intell Rev"},{"key":"R16","doi-asserted-by":"publisher","DOI":"10.3389\/fdgth.2025.1614105","article-title":"Biases in AI: acknowledging and addressing the inevitable ethical issues","volume":"7","author":"Hofmann","journal-title":"Front Digit Health"},{"issue":"1","key":"R17","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-025-03500-y","article-title":"Effect of delayed treatment on survival of patients with head and neck squamous cell cancer","volume":"15","author":"He","journal-title":"Sci Rep"},{"key":"R18","unstructured":"Lewis M Thio S Roberts A Siju C Mukit W Kuruvilla R et al. Grounding large language models in clinical evidence: a retrieval-augmented generation system for querying UK NICE clinical guidelines. arXiv. Preprint posted online on Oct 3, 2025. [doi: 10.48550\/arXiv.2510.02967]"},{"key":"R19","unstructured":"Lewis P Perez E Piktus A Petroni F Karpukhin V Goyal N et al. Retrieval-augmented generation for knowledge-intensive NLP tasks. In: Larochelle H Ranzato M Hadsell R Balcan MF Lin H , editors. NIPS \u201920: Proceedings of the 34th International Conference on Neural Information Processing Systems. Curran Associates Inc; 2020:9459-9474. URL: https:\/\/dl.acm.org\/doi\/abs\/10.5555\/3495724.3496517 [Accessed 28-07-2026]"},{"key":"R20","unstructured":"Zhang AL Kraska T Khattab O . Recursive language models. arXiv. Preprint posted online on Dec 31, 2025. [doi: 10.48550\/arXiv.2512.24601]"},{"issue":"1","key":"R21","doi-asserted-by":"publisher","DOI":"10.1038\/s41698-025-01157-4","article-title":"Real-world application of large language models for automated TNM staging using unstructured gynecologic oncology reports","volume":"9","author":"Ishida","journal-title":"NPJ Precis Oncol"},{"issue":"1","key":"R22","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-024-53081-z","article-title":"Matching patients to clinical trials with large language models","volume":"15","author":"Jin","journal-title":"Nat Commun"},{"issue":"1","key":"R23","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-024-01274-7","article-title":"PRISM: patient records interpretation for semantic clinical trial matching system using large language models","volume":"7","author":"Gupta","journal-title":"NPJ Digit Med"},{"key":"R24","doi-asserted-by":"publisher","DOI":"10.1016\/j.jbi.2024.104734","article-title":"Learning to match patients to clinical trials using large language models","volume":"159","author":"Rybinski","journal-title":"J Biomed Inform"}],"container-title":["JMIR AI"],"original-title":[],"language":"en","deposited":{"date-parts":[[2026,8,20]],"date-time":"2026-08-20T16:22:26Z","timestamp":1787242946000},"score":1,"resource":{"primary":{"URL":"https:\/\/ai.jmir.org\/2026\/1\/e95081"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,20]]},"references-count":24,"URL":"https:\/\/doi.org\/10.2196\/95081","relation":{},"ISSN":["2817-1705"],"issn-type":[{"value":"2817-1705","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,20]]},"article-number":"v5i4e95081"}}