{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,17]],"date-time":"2026-09-17T21:05:17Z","timestamp":1789679117289,"version":"4.0.1"},"reference-count":59,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["J Med Internet Res"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec sec-type=\"background\">\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Large language models (LLMs) demonstrate strong performance on medical knowledge benchmarks, but their safe and effective use in clinical practice depends on posttraining adaptation rather than raw model capability. Fine-tuning, retrieval-augmented generation (RAG), and hybrid approaches are principal strategies for grounding language models in clinical evidence, yet their comparative effectiveness remains unclear.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec sec-type=\"objective\">\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>This systematic review aims to synthesize evidence on fine-tuning, RAG, and hybrid posttraining strategies for clinical diagnosis and decision-support tasks and to identify strategy-task alignments and methodological features associated with improved performance.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec sec-type=\"methods\">\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>We conducted a systematic review in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines. PubMed\/MEDLINE, Scopus, and Web of Science were searched from January 2018 through May 2026. Eligible studies evaluated transformer-based language models that underwent posttraining adaptation, retrieval augmentation, or both for clinical decision support, diagnosis, triage, risk stratification, or related health care applications. Studies evaluating nonadapted models, non\u2013language-model AI systems, prompt engineering without performance evaluation, or nonclinical applications were excluded. Data extracted included model architecture, adaptation strategy, clinical domain, validation approach, and performance outcomes. Risk of bias was assessed using PROBAST+AI (Prediction model Risk of Bias Assessment Tool for AI). Studies were grouped according to the primary enhancement strategy (fine-tuning or parameter-efficient fine-tuning, RAG, or hybrid approaches), and findings were synthesized descriptively.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec sec-type=\"results\">\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Of 1890 identified records, 35 studies published between 2024 and 2026 met eligibility criteria. Enhancement strategies included RAG (17\/35, 48.6%), fine-tuning or parameter-efficient fine-tuning (7\/35, 20%), and hybrid approaches (11\/35, 31.4%). Studies included diverse specialties from oncology, neurology, radiology, mental health, cardiology, ophthalmology, and surgical care. Fine-tuning demonstrated strong performance for task-specific applications, achieving area under the receiver operating characteristic curve values up to 0.912 for cancer detection and area under curve of 0.892 for major depressive disorder prediction, while matching clinician-level diagnostic performance in several studies. RAG improved guideline adherence and diagnostic accuracy, with increases from 71.1% to 92.1% and from 78.9% to 94.7% in guideline-based decision-support tasks. However, benefits were inconsistent across larger reasoning-capable models. Hybrid systems generally achieved the strongest performance in complex clinical workflows, with external validation accuracies exceeding 90% in stroke triage, dermatology, multimodal imaging, and oncology applications. Risk-of-bias assessment identified substantial methodological limitations, with 25 studies judged as high risk, 9 as unclear risk, and only 1 as low risk overall. Common concerns included inadequate external validation, lack of calibration assessment, nonrepresentative participant selection, and insufficient reporting of analytical methods.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec sec-type=\"conclusions\">\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>Adaptation strategies should align with task needs, using fine-tuning for narrow classification, RAG for guideline-grounded reasoning, and hybrid approaches for complex multimodal tasks. However, the evidence base remains largely retrospective or benchmark-based. Prospective studies with external validation, calibration, and standardized safety reporting are needed before broader clinical use.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.2196\/104092","type":"journal-article","created":{"date-parts":[[2026,8,24]],"date-time":"2026-08-24T17:00:32Z","timestamp":1787590832000},"page":"e104092-e104092","source":"Crossref","is-referenced-by-count":0,"title":["Fine-Tuning, Retrieval-Augmented Generation, and Hybrid Adaptation of Language Models for Clinical Decision-Making in Health Care: Systematic Review"],"prefix":"10.2196","volume":"28","author":[{"given":"Anshum","family":"Patel","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3057-3200","authenticated-orcid":false,"given":"Yugant","family":"Khand","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sai Krishna","family":"Vallamchetla","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7015-0491","authenticated-orcid":false,"given":"Pengze","family":"Li","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4267-1924","authenticated-orcid":false,"given":"Cui","family":"Tao","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1222-7603","authenticated-orcid":false,"given":"Joseph","family":"Cheung","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,9,17]]},"reference":[{"issue":"14","key":"R1","doi-asserted-by":"publisher","first-page":"1347","DOI":"10.1056\/NEJMra1814259","article-title":"Machine learning in medicine","volume":"380","author":"Rajkomar","journal-title":"N Engl J Med"},{"issue":"1","key":"R2","doi-asserted-by":"publisher","first-page":"24","DOI":"10.1038\/s41591-018-0316-z","article-title":"A guide to deep learning in healthcare","volume":"25","author":"Esteva","journal-title":"Nat Med"},{"issue":"1","key":"R3","doi-asserted-by":"publisher","first-page":"44","DOI":"10.1038\/s41591-018-0300-7","article-title":"High-performance medicine: the convergence of human and artificial intelligence","volume":"25","author":"Topol","journal-title":"Nat Med"},{"issue":"7956","key":"R4","doi-asserted-by":"publisher","first-page":"259","DOI":"10.1038\/s41586-023-05881-4","article-title":"Foundation models for generalist medical artificial intelligence","volume":"616","author":"Moor","journal-title":"Nature"},{"issue":"9","key":"R5","doi-asserted-by":"publisher","first-page":"866","DOI":"10.1001\/jama.2023.14217","article-title":"Creation and adoption of large language models in medicine","volume":"330","author":"Shah","journal-title":"JAMA"},{"issue":"8","key":"R6","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":"7972","key":"R7","doi-asserted-by":"publisher","first-page":"172","DOI":"10.1038\/s41586-023-06291-2","article-title":"Large language models encode clinical knowledge","volume":"620","author":"Singhal","journal-title":"Nature"},{"issue":"6","key":"R8","doi-asserted-by":"publisher","first-page":"589","DOI":"10.1001\/jamainternmed.2023.1838","article-title":"Comparing physician and artificial intelligence chatbot responses to patient questions posted to a public social media forum","volume":"183","author":"Ayers","journal-title":"JAMA Intern Med"},{"issue":"4","key":"R9","doi-asserted-by":"publisher","first-page":"1134","DOI":"10.1038\/s41591-024-02855-5","article-title":"Adapted large language models can outperform medical experts in clinical text summarization","volume":"30","author":"Van Veen","journal-title":"Nat Med"},{"issue":"3","key":"R10","doi-asserted-by":"publisher","first-page":"943","DOI":"10.1038\/s41591-024-03423-7","article-title":"Toward expert-level medical question answering with large language models","volume":"31","author":"Singhal","journal-title":"Nat Med"},{"issue":"3","key":"R11","doi-asserted-by":"publisher","first-page":"1152","DOI":"10.1038\/s41591-026-04229-5","article-title":"LLM-assisted systematic review of large language models in clinical medicine","volume":"32","author":"Chen","journal-title":"Nat Med"},{"issue":"1","key":"R12","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01963-x","article-title":"The evaluation illusion of large language models in medicine","volume":"8","author":"Agrawal","journal-title":"NPJ Digit Med"},{"issue":"9","key":"R13","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"},{"issue":"1","key":"R14","doi-asserted-by":"publisher","first-page":"78","DOI":"10.1001\/jama.2023.8288","article-title":"Accuracy of a generative artificial intelligence model in a complex diagnostic challenge","volume":"330","author":"Kanjee","journal-title":"JAMA"},{"issue":"5","key":"R15","doi-asserted-by":"publisher","first-page":"581","DOI":"10.1001\/jamainternmed.2024.0295","article-title":"Clinical reasoning of a generative artificial intelligence model compared with physicians","volume":"184","author":"Cabral","journal-title":"JAMA Intern Med"},{"issue":"10","key":"R16","doi-asserted-by":"publisher","DOI":"10.1001\/jamanetworkopen.2024.40969","article-title":"Large language model influence on diagnostic reasoning: a randomized clinical trial","volume":"7","author":"Goh","journal-title":"JAMA Netw Open"},{"issue":"1","key":"R17","doi-asserted-by":"publisher","DOI":"10.1038\/s43856-025-01021-3","article-title":"Multi-model assurance analysis showing large language models are highly vulnerable to adversarial hallucination attacks during clinical decision support","volume":"5","author":"Omar","journal-title":"Commun Med (Lond)"},{"issue":"1","key":"R18","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-023-00939-z","article-title":"Large language models propagate race-based medicine","volume":"6","author":"Omiye","journal-title":"NPJ Digit Med"},{"issue":"1","key":"R19","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-024-01239-w","article-title":"Closing the gap between open source and commercial large language models for medical evidence summarization","volume":"7","author":"Zhang","journal-title":"NPJ Digit Med"},{"issue":"1","key":"R20","doi-asserted-by":"publisher","DOI":"10.1038\/s44401-024-00004-1","article-title":"Retrieval-augmented generation for generative artificial intelligence in health care","volume":"2","author":"Yang","journal-title":"Npj Health Syst"},{"issue":"1","key":"R21","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01802-z","article-title":"Retrieval-augmented generation elevates local LLM quality in radiology contrast media consultation","volume":"8","author":"Wada","journal-title":"NPJ Digit Med"},{"key":"R22","doi-asserted-by":"publisher","DOI":"10.1136\/bmj.n71","article-title":"The PRISMA 2020 statement: an updated guideline for reporting systematic reviews","volume":"372","author":"Page","journal-title":"BMJ"},{"key":"R23","doi-asserted-by":"publisher","DOI":"10.1136\/bmj-2024-082505","article-title":"PROBAST+AI: an updated quality, risk of bias, and applicability assessment tool for prediction models using regression or artificial intelligence methods","volume":"388","author":"Moons","journal-title":"BMJ"},{"issue":"5","key":"R24","doi-asserted-by":"publisher","DOI":"10.1093\/bib\/bbae430","article-title":"Large language model produces high accurate diagnosis of cancer from end-motif profiles of cell-free DNA","volume":"25","author":"Liu","journal-title":"Brief Bioinform"},{"issue":"1","key":"R25","doi-asserted-by":"publisher","DOI":"10.1007\/s10916-025-02289-7","article-title":"Estimating LVEF from ECG with GPT-4o fine-tuned vision: a novel approach in AI-driven cardiac diagnostics","volume":"49","author":"Engelstein","journal-title":"J Med Syst"},{"issue":"4","key":"R26","doi-asserted-by":"publisher","first-page":"495","DOI":"10.1007\/s10140-025-02354-1","article-title":"Fine-tuned large language model for extracting newly identified acute brain infarcts based on computed tomography or magnetic resonance imaging reports","volume":"32","author":"Fujita","journal-title":"Emerg Radiol"},{"key":"R27","doi-asserted-by":"publisher","DOI":"10.1016\/j.jbi.2025.104830","article-title":"CD-Tron: leveraging large clinical language model for early detection of cognitive decline from electronic health records","volume":"166","author":"Guan","journal-title":"J Biomed Inform"},{"issue":"7","key":"R28","doi-asserted-by":"publisher","DOI":"10.7759\/cureus.88711","article-title":"Multiclass classification of pigmented skin lesions using a multimodal large language model","volume":"17","author":"Iinuma","journal-title":"Cureus"},{"key":"R29","doi-asserted-by":"publisher","DOI":"10.2196\/65263","article-title":"Large language models for pediatric differential diagnoses in rural health care: multicenter retrospective cohort study comparing GPT-3 with pediatrician performance","volume":"6","author":"Mansoor","journal-title":"JMIRx Med"},{"key":"R30","doi-asserted-by":"publisher","DOI":"10.1016\/j.jad.2025.119774","article-title":"MDD-LLM: towards accuracy large language models for major depressive disorder diagnosis","volume":"388","author":"Sha","journal-title":"J Affect Disord"},{"key":"R31","doi-asserted-by":"publisher","DOI":"10.1200\/PO-24-00478","article-title":"Expert-guided large language models for clinical decision support in precision oncology","volume":"8","author":"Lammert","journal-title":"JCO Precis Oncol"},{"issue":"1","key":"R32","doi-asserted-by":"publisher","DOI":"10.1186\/s41747-024-00457-x","article-title":"A retrieval-augmented chatbot based on GPT-4 provides appropriate differential diagnosis in gastrointestinal radiology: a proof of concept study","volume":"8","author":"Rau","journal-title":"Eur Radiol Exp"},{"key":"R33","doi-asserted-by":"publisher","DOI":"10.1016\/j.jbi.2024.104702","article-title":"Rare disease diagnosis using knowledge guided retrieval augmentation for ChatGPT","volume":"157","author":"Zelin","journal-title":"J Biomed Inform"},{"issue":"20","key":"R34","doi-asserted-by":"publisher","first-page":"2458","DOI":"10.1016\/j.jcin.2025.08.019","article-title":"Performance of large language models on the acute coronary syndrome guidelines using retrieval-augmented generation","volume":"18","author":"Alexandrou","journal-title":"JACC Cardiovasc Interv"},{"key":"R35","doi-asserted-by":"publisher","DOI":"10.1101\/2025.07.03.25330805","article-title":"GlaucoRAG: a retrieval-augmented large language model for expert-level glaucoma assessment","author":"Aminan","journal-title":"medRxiv"},{"key":"R36","doi-asserted-by":"publisher","DOI":"10.3389\/fmed.2025.1599241","article-title":"Enhancing treatment decision-making for low back pain: a novel framework integrating large language models with retrieval-augmented generation technology","volume":"12","author":"Chen","journal-title":"Front Med"},{"key":"R37","doi-asserted-by":"publisher","DOI":"10.1016\/j.eswa.2025.127191","article-title":"Decoding the mind: a RAG-LLM on ICD-11 for decision support in psychology","volume":"279","author":"Cremaschi","journal-title":"Expert Syst Appl"},{"issue":"8","key":"R38","doi-asserted-by":"publisher","first-page":"5091","DOI":"10.1007\/s00330-025-11445-z","article-title":"Retrieval-augmented generation improves precision and trust of a GPT-4 model for emergency radiology diagnosis and classification: a proof-of-concept study","volume":"35","author":"Fink","journal-title":"Eur Radiol"},{"issue":"1","key":"R39","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01684-1","article-title":"Evaluating large language model workflows in clinical decision support for triage and referral and diagnosis","volume":"8","author":"Gaber","journal-title":"NPJ Digit Med"},{"key":"R40","doi-asserted-by":"publisher","DOI":"10.1109\/EMBC58623.2025.11253463","article-title":"Retrieval-augmented generation for medical decision-making in emergency care","volume":"2025","author":"Noll","journal-title":"Annu Int Conf IEEE Eng Med Biol Soc"},{"issue":"10","key":"R41","doi-asserted-by":"publisher","DOI":"10.1016\/j.xcrm.2025.102323","article-title":"Large language model as clinical decision support system augments medication safety in 16 clinical specialties","volume":"6","author":"Ong","journal-title":"Cell Rep Med"},{"issue":"4","key":"R42","doi-asserted-by":"publisher","DOI":"10.1016\/j.jham.2025.100293","article-title":"Development of a novel artificial intelligence clinical decision support tool for hand surgery: HandRAG","volume":"17","author":"Ozmen","journal-title":"J Hand Microsurg"},{"issue":"8","key":"R43","doi-asserted-by":"publisher","DOI":"10.1002\/micr.70138","article-title":"MicroRAG: development of a novel artificial intelligence retrieval-augmented generation model for microsurgery clinical decision support","volume":"45","author":"Ozmen","journal-title":"Microsurgery"},{"key":"R44","doi-asserted-by":"publisher","DOI":"10.1200\/CCI-25-00220","article-title":"RadOncRAG: a novel retrieval-augmented generation framework improves large language model benchmark performance in radiation oncology","volume":"9","author":"Thaker","journal-title":"JCO Clin Cancer Inform"},{"key":"R45","doi-asserted-by":"publisher","DOI":"10.2196\/78393","article-title":"Performance of retrieval-augmented generation large language models in guideline-concordant prostate-specific antigen testing: comparative study with junior clinicians","volume":"27","author":"Tung","journal-title":"J Med Internet Res"},{"issue":"1","key":"R46","doi-asserted-by":"publisher","DOI":"10.1016\/j.identj.2025.109344","article-title":"Evaluating retrieval-augmented generation-large language models for infective endocarditis prophylaxis: clinical accuracy and efficiency","volume":"76","author":"Rewthamrongsris","journal-title":"Int Dent J"},{"issue":"8","key":"R47","doi-asserted-by":"publisher","first-page":"1436","DOI":"10.1093\/jamia\/ocag056","article-title":"Optimizing retrieval-augmented generation (RAG) in clinical medicine: methods and performance evaluation","volume":"33","author":"Li","journal-title":"J Am Med Inform Assoc"},{"issue":"12","key":"R48","doi-asserted-by":"publisher","first-page":"1324","DOI":"10.1177\/00220345251334575","article-title":"Multimodal language model for jaw osteonecrosis diagnosis and treatment","volume":"104","author":"Gao","journal-title":"J Dent Res"},{"issue":"1","key":"R49","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-02139-3","article-title":"Enhancing privacy-preserving deployable large language models for perioperative complication detection: a targeted strategy with LoRA fine-tuning","volume":"8","author":"Gao","journal-title":"NPJ Digit Med"},{"issue":"1","key":"R50","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-025-27410-1","article-title":"Ophtimus-V2-Tx: a compact domain-specific LLM for ophthalmic diagnosis and treatment planning","volume":"15","author":"Kwon","journal-title":"Sci Rep"},{"issue":"3","key":"R51","doi-asserted-by":"publisher","first-page":"89","DOI":"10.3390\/make7030089","article-title":"AlzheimerRAG: multimodal retrieval-augmented generation for clinical use cases","volume":"7","author":"Lahiri","journal-title":"Mach Learn Knowl Extr"},{"issue":"12","key":"R52","doi-asserted-by":"publisher","DOI":"10.1002\/alz.71029","article-title":"A vision\u2013language foundation model for Alzheimer\u2019s disease diagnosis using MRI and clinical data","volume":"21","author":"Lin","journal-title":"Alzheimers Dement"},{"key":"R53","doi-asserted-by":"publisher","DOI":"10.2196\/67010","article-title":"Stroke diagnosis and prediction tool using ChatGLM: development and validation study","volume":"27","author":"Song","journal-title":"J Med Internet Res"},{"issue":"9","key":"R54","doi-asserted-by":"publisher","DOI":"10.1088\/1361-6560\/adcb17","article-title":"Diagnosis assistant for liver cancer utilizing a large language model with three types of knowledge","volume":"70","author":"Wu","journal-title":"Phys Med Biol"},{"issue":"7","key":"R55","doi-asserted-by":"publisher","DOI":"10.3390\/bioengineering12070698","article-title":"Adaptive RAG-assisted MRI platform (ARAMP) for brain metastasis detection and reporting: a retrospective evaluation using post-contrast T1-weighted imaging","volume":"12","author":"Wu","journal-title":"Bioengineering (Basel)"},{"issue":"2","key":"R56","doi-asserted-by":"publisher","first-page":"211","DOI":"10.1007\/s11548-025-03524-9","article-title":"Large language models with retrieval-augmented generation enhance expert modelling of Bayesian network for clinical decision support","volume":"21","author":"Cypko","journal-title":"Int J Comput Assist Radiol Surg"},{"issue":"1","key":"R57","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-026-38715-0","article-title":"DermaGPT a federated multimodal framework with a meta learned trust function for interpretable dermatology diagnostics","volume":"16","author":"Hashjin","journal-title":"Sci Rep"},{"key":"R58","doi-asserted-by":"publisher","DOI":"10.2196\/71541","article-title":"Personalized diabetes treatment support using large language models fine-tuned on electronic health records: development and evaluation study","volume":"10","author":"He","journal-title":"JMIR Form Res"},{"issue":"1","key":"R59","doi-asserted-by":"publisher","DOI":"10.1016\/j.mcpdig.2024.11.005","article-title":"Fine-tuning large language models for specialized use cases","volume":"3","author":"Anisuzzaman","journal-title":"Mayo Clin Proc Digit Health"}],"container-title":["Journal of Medical Internet Research"],"original-title":[],"language":"en","deposited":{"date-parts":[[2026,9,17]],"date-time":"2026-09-17T20:41:14Z","timestamp":1789677674000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.jmir.org\/2026\/1\/e104092"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,17]]},"references-count":59,"URL":"https:\/\/doi.org\/10.2196\/104092","relation":{},"ISSN":["1438-8871"],"issn-type":[{"value":"1438-8871","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,17]]},"article-number":"v28i14e104092"}}