{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,4]],"date-time":"2026-08-04T19:51:18Z","timestamp":1785873078830,"version":"3.56.0"},"reference-count":31,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["JMIR Med Inform"],"abstract":"Abstract<\/jats:title>\n \n Background<\/jats:title>\n Pneumonia is a common infectious disease, and antibiotic treatment in hospitalized patients must balance efficacy, safety, and resistance risk. However, antibiotic selection and dose adjustment still rely heavily on clinician experience. Although large language models (LLMs) are promising for clinical reasoning, their direct use for antibiotic selection and dose recommendation is limited by hallucinations and weak adherence to clinical constraints.<\/jats:p>\n <\/jats:sec>\n \n Objective<\/jats:title>\n This study aimed to develop and externally validate a constrained LLM-based clinical decision support pipeline for antibiotic selection and dose recommendation in hospitalized patients with pneumonia.<\/jats:p>\n <\/jats:sec>\n \n Methods<\/jats:title>\n \n We conducted a multicenter retrospective study using electronic health record narratives, antibiotic orders, and laboratory indicators of hepatic and renal function from 331 hospitalized patients with pneumonia from 2 hospitals in China. The development cohort included 233 patients, and the external validation cohort included 98 patients. The pipeline integrated dual-branch retrieval (similar-case vector retrieval plus guideline-based knowledge graph retrieval), clinician-defined rule constraints, and hybrid-context reasoning. DeepSeek-V3, GLM-4.6, and GPT-4o were evaluated using\n F<\/jats:italic>\n 1<\/jats:sub>\n -score and Jaccard accuracy.\n <\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n \n On the internal test set, the full pipeline using DeepSeek-V3 achieved the best performance, with an\n F<\/jats:italic>\n 1<\/jats:sub>\n -score of 0.8110 (95% CI 0.7371-0.8762) and Jaccard accuracy of 0.7624 (95% CI 0.6810-0.8386) for antibiotic selection and an\n F<\/jats:italic>\n 1<\/jats:sub>\n -score of 0.7538 (95% CI 0.6671-0.8329) and Jaccard accuracy of 0.7076 (95% CI 0.6145-0.7938) for joint antibiotic selection plus dosing recommendation. On the external validation set, performance remained high, with an\n F<\/jats:italic>\n 1<\/jats:sub>\n -score of 0.8605 (95% CI 0.7891-0.9252) and Jaccard accuracy of 0.8571 (95% CI 0.7857-0.9184) for antibiotic selection, and an\n F<\/jats:italic>\n 1<\/jats:sub>\n -score of 0.8503 (95% CI 0.7789-0.9150) and Jaccard accuracy of 0.8469 (95% CI 0.7755-0.9133) for antibiotic selection plus dosing recommendation. The system also provided traceable evidence and rule trigger information to support clinician review.\n <\/jats:p>\n <\/jats:sec>\n \n Conclusions<\/jats:title>\n A constrained, retrieval-augmented LLM pipeline improved the consistency and interpretability of antibiotic selection and dose recommendation for hospitalized patients with pneumonia and provided preliminary evidence of cross-site generalizability.<\/jats:p>\n <\/jats:sec>","DOI":"10.2196\/98207","type":"journal-article","created":{"date-parts":[[2026,8,4]],"date-time":"2026-08-04T19:41:22Z","timestamp":1785872482000},"page":"e98207-e98207","source":"Crossref","is-referenced-by-count":0,"title":["Large Language Model\u2013Based Clinical Decision Support for Antibiotic Selection and Dose Recommendation in Hospitalized Patients With Pneumonia: Multicenter Retrospective Study"],"prefix":"10.2196","volume":"14","author":[{"given":"Yang","family":"Zhang","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Li","family":"Li","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6004-3390","authenticated-orcid":false,"given":"Chunting","family":"Tan","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mengyuan","family":"Ji","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xican","family":"Tian","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiangdong","family":"Mu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jun","family":"Li","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yu","family":"Gu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5518-4749","authenticated-orcid":false,"given":"Honglei","family":"Liu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,8,4]]},"reference":[{"key":"R1","doi-asserted-by":"publisher","DOI":"10.1016\/j.lanwpc.2023.100968","article-title":"The mortality and years of life lost for community-acquired pneumonia before and during COVID-19 pandemic in China","volume":"42","author":"Fan","journal-title":"Lancet Reg Health West Pac"},{"issue":"10442","key":"R2","doi-asserted-by":"publisher","DOI":"10.1016\/S0140-6736(24)01076-6","article-title":"Antimicrobial resistance: an agenda for all","volume":"403","author":"The Lancet","journal-title":"Lancet"},{"issue":"10459","key":"R3","doi-asserted-by":"publisher","first-page":"1199","DOI":"10.1016\/S0140-6736(24)01867-1","article-title":"Global burden of bacterial antimicrobial resistance 1990-2021: a systematic analysis with forecasts to 2050","volume":"404","author":"GBD 2021 Antimicrobial Resistance Collaborators","journal-title":"Lancet"},{"key":"R4","volume":"5","author":"Shortliffe","journal-title":"Proc Annu Symp Comput Appl Med Care"},{"issue":"8","key":"R5","doi-asserted-by":"publisher","first-page":"468","DOI":"10.1056\/NEJM198208193070803","article-title":"Internist-1, an experimental computer-based diagnostic consultant for general internal medicine","volume":"307","author":"Miller","journal-title":"N Engl J Med"},{"issue":"1","key":"R6","doi-asserted-by":"publisher","first-page":"19","DOI":"10.1197\/jamia.M3378","article-title":"MedEx: a medication information extraction system for clinical narratives","volume":"17","author":"Xu","journal-title":"J Am Med Inform Assoc"},{"issue":"5","key":"R7","doi-asserted-by":"publisher","first-page":"507","DOI":"10.1136\/jamia.2009.001560","article-title":"Mayo clinical Text Analysis and Knowledge Extraction System (cTAKES): architecture, component evaluation and applications","volume":"17","author":"Savova","journal-title":"J Am Med Inform Assoc"},{"key":"R8","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-018-0029-1","article-title":"Scalable and accurate deep learning with electronic health records","volume":"1","author":"Rajkomar","journal-title":"NPJ Digit Med"},{"issue":"9","key":"R9","doi-asserted-by":"publisher","first-page":"517","DOI":"10.3390\/info15090517","article-title":"Recurrent neural networks: a comprehensive review of architectures, variants, and applications","volume":"15","author":"Mienye","journal-title":"Information"},{"issue":"10","key":"R10","doi-asserted-by":"publisher","first-page":"1419","DOI":"10.1093\/jamia\/ocy068","article-title":"Opportunities and challenges in developing deep learning models using electronic health records data: a systematic review","volume":"25","author":"Xiao","journal-title":"J Am Med Inform Assoc"},{"issue":"4","key":"R11","doi-asserted-by":"publisher","first-page":"1234","DOI":"10.1093\/bioinformatics\/btz682","article-title":"BioBERT: a pre-trained biomedical language representation model for biomedical text mining","volume":"36","author":"Lee","journal-title":"Bioinformatics"},{"key":"R12","unstructured":"Huang K Altosaar J Ranganath R . ClinicalBERT: modeling clinical notes and predicting hospital readmission. arXiv. Preprint posted online on Apr 10, 2019. [doi: 10.48550\/arXiv.1904.05342]"},{"issue":"1","key":"R13","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-021-00455-y","article-title":"Med-BERT: pretrained contextualized embeddings on large-scale structured electronic health records for disease prediction","volume":"4","author":"Rasmy","journal-title":"NPJ Digit Med"},{"issue":"3","key":"R14","doi-asserted-by":"publisher","first-page":"457","DOI":"10.1093\/jamia\/ocz200","article-title":"Deep learning in clinical natural language processing: a methodical review","volume":"27","author":"Wu","journal-title":"J Am Med Inform Assoc"},{"issue":"1","key":"R15","doi-asserted-by":"publisher","first-page":"15","DOI":"10.1002\/hast.973","article-title":"Artificial intelligence and black-box medical decisions: accuracy versus explainability","volume":"49","author":"London","journal-title":"Hastings Cent Rep"},{"issue":"11","key":"R16","doi-asserted-by":"publisher","first-page":"1688","DOI":"10.1093\/jamia\/ocaa154","article-title":"A machine learning-based clinical decision support system to identify prescriptions with a high risk of medication error","volume":"27","author":"Corny","journal-title":"J Am Med Inform Assoc"},{"issue":"1","key":"R17","doi-asserted-by":"publisher","first-page":"1126","DOI":"10.1609\/aaai.v33i01.33011126","article-title":"GAMENet: Graph Augmented MEmory Networks for recommending medication combination","volume":"33","author":"Shang","journal-title":"Proc AAAI Conf Artif Intell"},{"issue":"8","key":"R18","doi-asserted-by":"publisher","first-page":"1930","DOI":"10.1038\/s41591-023-02448-8","article-title":"Large language models in medicine","volume":"29","author":"Thirunavukarasu","journal-title":"Nat Med"},{"issue":"1","key":"R19","doi-asserted-by":"publisher","first-page":"100931","DOI":"10.1016\/j.landig.2025.100931","article-title":"Reasoning-driven large language models in medicine: opportunities, challenges, and the road ahead","volume":"8","author":"Wang","journal-title":"Lancet Digit Health"},{"key":"R20","unstructured":"Adams L Hu S Derby N et al. Large language models with temporal reasoning for longitudinal clinical summarization and prediction. arXiv. Preprint posted online on Jan 30, 2025. [doi: 10.48550\/arXiv.2501.18724]"},{"key":"R21","unstructured":"Boll HO Boll AO Boll LP Hanna AA Calixto I . DistillNote: LLM-based clinical note summaries improve heart failure diagnosis. arXiv. Preprint posted online on Jun 20, 2025. URL: https:\/\/arxiv.org\/html\/2506.16777v1 [Accessed 22-07-2026]"},{"issue":"1","key":"R22","doi-asserted-by":"publisher","DOI":"10.1136\/bmjhci-2024-101139","article-title":"Large language models for data extraction from unstructured and semi-structured electronic health records: a multiple model performance evaluation","volume":"32","author":"Ntinopoulos","journal-title":"BMJ Health Care Inform"},{"issue":"8","key":"R23","doi-asserted-by":"publisher","first-page":"1336","DOI":"10.1016\/j.cmi.2025.03.002","article-title":"Comparing large language models for antibiotic prescribing in different clinical scenarios: which performs better?","volume":"31","author":"De Vito","journal-title":"Clin Microbiol Infect"},{"key":"R24","unstructured":"Liu Q Wu X Zhao X et al. Large language model distilling medication recommendation model. arXiv. Preprint posted online on Feb 5, 2024. [doi: 10.48550\/arXiv.2402.02803]"},{"key":"R25","doi-asserted-by":"crossref","unstructured":"Li L Zhang Y Liu D Chen L . Large language models for generative recommendation: a survey and visionary discussions. arXiv. Preprint posted online on Sep 3, 2023. [doi: 10.48550\/arXiv.2309.01157]","DOI":"10.63317\/4pmv5q3r3yc5"},{"key":"R26","doi-asserted-by":"publisher","DOI":"10.3389\/fphar.2025.1514445","article-title":"Large language models management of complex medication regimens: a case-based evaluation","volume":"16","author":"Chase","journal-title":"Front Pharmacol"},{"issue":"4","key":"R27","doi-asserted-by":"publisher","first-page":"605","DOI":"10.1093\/jamia\/ocaf008","article-title":"Improving large language model applications in biomedicine with retrieval-augmented generation: a systematic review, meta-analysis, and clinical development guidelines","volume":"32","author":"Liu","journal-title":"J Am Med Inform Assoc"},{"key":"R28","doi-asserted-by":"publisher","DOI":"10.2196\/80557","article-title":"Improving large language model applications in the medical and nursing domains with retrieval-augmented generation: scoping review","volume":"27","author":"Miao","journal-title":"J Med Internet Res"},{"key":"R29","doi-asserted-by":"publisher","DOI":"10.2196\/82026","article-title":"Multi-evidence clinical reasoning with retrieval-augmented generation for emergency triage: retrospective evaluation study","volume":"14","author":"Wong","journal-title":"JMIR Med Inform"},{"key":"R30","unstructured":"Normand O Borsi E Fruin M et al. A real-world evaluation of LLM medication safety reviews in NHS primary care. arXiv. Preprint posted online on Dec 24, 2025. [doi: 10.48550\/arXiv.2512.21127]"},{"issue":"4","key":"R31","doi-asserted-by":"publisher","first-page":"241","DOI":"10.3760\/cma.j.issn.1001-0939.2016.04.001","article-title":"Guidelines for the diagnosis and treatment of adult community acquired pneumonia in China (2016 Edition) [Article in Chinese]","volume":"39","author":"Qu","journal-title":"Zhonghua Jie He He Hu Xi Za Zhi"}],"container-title":["JMIR Medical Informatics"],"original-title":[],"language":"en","deposited":{"date-parts":[[2026,8,4]],"date-time":"2026-08-04T19:41:26Z","timestamp":1785872486000},"score":1,"resource":{"primary":{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e98207"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,4]]},"references-count":31,"URL":"https:\/\/doi.org\/10.2196\/98207","relation":{},"ISSN":["2291-9694"],"issn-type":[{"value":"2291-9694","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,4]]},"article-number":"v14i10442e98207"}}