{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,10]],"date-time":"2026-09-10T22:09:54Z","timestamp":1789078194450,"version":"build-2803163510"},"reference-count":47,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["JMIR Res Protoc"],"abstract":"Abstract<\/jats:title>\n \n Background<\/jats:title>\n Large language models (LLMs) are increasingly applied in health care for clinical decision-making, education, and patient communication. However, bias in LLM outputs may exacerbate health care disparities and compromise trust. Despite rapid adoption, there is limited synthesis of how bias is identified, measured, and mitigated in medical applications of LLMs.<\/jats:p>\n <\/jats:sec>\n \n Objective<\/jats:title>\n This systematic review aims to evaluate how bias is detected, measured, and mitigated in health care applications of LLMs and to identify methodological trends and gaps in the current literature.<\/jats:p>\n <\/jats:sec>\n \n Methods<\/jats:title>\n We will conduct a systematic review of studies evaluating bias, fairness, or subgroup performance in medical applications of LLMs. Searches will be conducted across Embase, MEDLINE, PsycINFO, PubMed, ACL Anthology, ACM Digital Library, arXiv, medRxiv, and bioRxiv for studies published from 2017 onward. The review uses a staged design. In phase 1, records identified in the original June 2025 search were screened and assessed manually using predefined eligibility criteria. This fully manually reviewed dataset will serve as a reference standard for validating the review workflow. In phase 2, an updated June 2026 search will be screened using a prespecified LLM-assisted workflow. Two independent LLMs will apply the same eligibility criteria used in manual screening, with studies marked as potentially relevant or uncertain by either model progressing to further assessment. Final inclusion decisions will be made by at least 2 human reviewers. The performance of the LLM-assisted workflow will be evaluated against the manually reviewed reference dataset using sensitivity, specificity, precision, negative predictive value, inclusion agreement, and Cohen \u03ba. Data will be extracted using a standardized framework covering study characteristics, model details, health care use case, bias type, bias assessment methods, mitigation strategies, transparency, generalizability, and ethical or regulatory framing. Findings will be synthesized narratively.<\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n The phase 1 search, conducted on June 11, 2025, yielded 15,976 records after deduplication and has undergone complete manual screening. The phase 2 search, conducted on June 8, 2026, increased the total number of records to 41,143. Screening of the updated dataset using the LLM-assisted workflow is ongoing. The completed review will report included study characteristics, approaches to bias detection and mitigation, and validation metrics for the LLM-assisted workflow. The review is expected to be submitted for publication in late November 2026, with publication anticipated thereafter subject to the journal\u2019s peer-review and editorial process.<\/jats:p>\n <\/jats:sec>\n \n Conclusions<\/jats:title>\n This review aims to provide a comprehensive synthesis of current approaches to bias detection and mitigation in medical LLMs, highlighting methodological strengths, limitations, and areas for future research. The findings aim to inform the development of more equitable and transparent AI systems in health care.<\/jats:p>\n <\/jats:sec>","DOI":"10.2196\/91348","type":"journal-article","created":{"date-parts":[[2026,7,30]],"date-time":"2026-07-30T09:05:06Z","timestamp":1785402306000},"page":"e91348-e91348","source":"Crossref","is-referenced-by-count":0,"title":["Exploring Bias in Medical Applications of Large Language Models: Protocol for a Systematic Review"],"prefix":"10.2196","volume":"15","author":[{"ORCID":"https:\/\/orcid.org\/0009-0006-3483-0610","authenticated-orcid":false,"given":"Shweta","family":"Rao","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0003-1527-2375","authenticated-orcid":false,"given":"Carol","family":"Boutrous","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Andrey","family":"Kormilitzin","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7333-2204","authenticated-orcid":false,"given":"Xin You","family":"Tai","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1878-3264","authenticated-orcid":false,"given":"Lewis","family":"Hotchkiss","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-8261-6480","authenticated-orcid":false,"given":"Cen","family":"Cong","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2484-3347","authenticated-orcid":false,"given":"Edward","family":"Meinert","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4408-4528","authenticated-orcid":false,"given":"Huizhi","family":"Liang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6828-6891","authenticated-orcid":false,"given":"Hang","family":"Dong","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5775-2524","authenticated-orcid":false,"given":"Judith R","family":"Harrison","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,9,10]]},"reference":[{"key":"R1","unstructured":"Radford A Narasimhan K Salimans T Sutskever I . Improving language understanding by generative pre-training. OpenAI. 2018. URL: https:\/\/openai.com\/index\/language-unsupervised\/ [Accessed 10-01-2026]"},{"key":"R2","doi-asserted-by":"crossref","unstructured":"Devlin J Chang MW Lee K Toutanova K . BERT: pre-training of deep bidirectional transformers for language understanding. Presented at: 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT 2019); Jun 2-7, 2019; Minneapolis, Minnesota, USA. URL: https:\/\/aclanthology.org\/N19-1423.pdf [Accessed 08-08-2026]","DOI":"10.18653\/v1\/N19-1423"},{"key":"R3","doi-asserted-by":"publisher","DOI":"10.5555\/3648699.3648939","article-title":"PaLM: scaling language modeling with pathways","volume":"24","author":"Chowdhery","journal-title":"J Mach Learn Res"},{"key":"R4","unstructured":"Touvron H Lavril T Izacard G et al. LLaMA: open and efficient foundation language models. arXiv. Preprint posted online on Feb 27, 2023. [doi: 10.48550\/arXiv.2302.13971]"},{"issue":"11","key":"R5","doi-asserted-by":"publisher","first-page":"299","DOI":"10.1007\/s10462-024-10921-0","article-title":"Large language models in medical and healthcare fields: applications, advances, and challenges","volume":"57","author":"Wang","journal-title":"Artif Intell Rev"},{"issue":"5","key":"R6","doi-asserted-by":"publisher","DOI":"10.1016\/j.isci.2024.109713","article-title":"The application of large language models in medicine: a scoping review","volume":"27","author":"Meng","journal-title":"iScience"},{"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"},{"key":"R8","doi-asserted-by":"publisher","DOI":"10.2196\/59069","article-title":"Revolutionizing health care: the transformative impact of large language models in medicine","volume":"27","author":"Zhang","journal-title":"J Med Internet Res"},{"key":"R9","doi-asserted-by":"publisher","DOI":"10.1016\/j.ceh.2024.12.001","article-title":"Conversational AI with large language models to increase the uptake of clinical guidance","volume":"7","author":"Macia","journal-title":"Clin eHealth"},{"issue":"4","key":"R10","doi-asserted-by":"publisher","first-page":"88","DOI":"10.1007\/s13347-022-00577-5","article-title":"AI, opacity, and personal autonomy","volume":"35","author":"Vaassen","journal-title":"Philos Technol"},{"key":"R11","unstructured":"Li S Rui H . Dual traits in probabilistic reasoning of large language models. arXiv. Preprint posted online on Dec 15, 2024. [doi: 10.48550\/arXiv.2412.11009]"},{"issue":"1","key":"R12","doi-asserted-by":"publisher","first-page":"52","DOI":"10.1016\/j.imed.2023.08.001","article-title":"Medical artificial intelligence and the black box problem: a view based on the ethical principle of \u201cdo no harm\u201d","volume":"4","author":"Xu","journal-title":"Intell Med"},{"issue":"6","key":"R13","doi-asserted-by":"publisher","first-page":"e428","DOI":"10.1016\/S2589-7500(24)00061-X","article-title":"Ethical and regulatory challenges of large language models in medicine","volume":"6","author":"Ong","journal-title":"Lancet Digit Health"},{"issue":"9","key":"R14","doi-asserted-by":"publisher","first-page":"1824","DOI":"10.3390\/electronics15091824","article-title":"Bias in large language models: origin, evaluation, and mitigation","volume":"15","author":"Guo","journal-title":"Electronics (Basel)"},{"issue":"2","key":"R15","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3703155","article-title":"A survey on hallucination in large language models: principles, taxonomy, challenges, and open questions","volume":"43","author":"Huang","journal-title":"ACM Trans Inf Syst"},{"issue":"1","key":"R16","doi-asserted-by":"publisher","first-page":"385","DOI":"10.1146\/annurev-med-042921-124419","article-title":"Use of race in kidney function estimation: lessons learned and the path toward health justice","volume":"74","author":"Mohottige","journal-title":"Annu Rev Med"},{"issue":"22","key":"R17","doi-asserted-by":"publisher","DOI":"10.1136\/heartjnl-2024-324625","article-title":"Advancing the access to cardiovascular diagnosis and treatment among women with cardiovascular disease: a joint British Cardiovascular Societies\u2019 consensus document","volume":"110","author":"Tayal","journal-title":"Heart"},{"key":"R18","doi-asserted-by":"crossref","unstructured":"Gallegos IO Rossi RA Barrow J et al. Bias and fairness in large language models: a survey. In: Computational Linguistics. MIT Press; 2024:1097-1179. [doi: 10.1162\/coli_a_00524]","DOI":"10.1162\/coli_a_00524"},{"key":"R19","unstructured":"Chaudhary I Hu Q Kumar M Ziyadi M Gupta R Singh G . Certifying counterfactual bias in LLMs. Presented at: The Thirteenth International Conference on Learning Representations (ICLR 2025); Apr 24-28, 2025; Singapore. URL: https:\/\/openreview.net\/forum?id=HQHnhVQznF [Accessed 08-08-2026]"},{"issue":"2","key":"R20","doi-asserted-by":"publisher","first-page":"1329","DOI":"10.32628\/CSEIT25112461","article-title":"Bias detection and fairness in large language models for financial services","volume":"11","author":"Vats","journal-title":"Int J Sci Res Comput Sci Eng Inf Technol"},{"issue":"4","key":"R21","doi-asserted-by":"publisher","DOI":"10.1016\/j.patter.2024.100946","article-title":"An evaluation of synthetic data augmentation for mitigating covariate bias in health data","volume":"5","author":"Juwara","journal-title":"Patterns (N Y)"},{"key":"R22","doi-asserted-by":"crossref","unstructured":"Hu M Wu H Zhu R et al. No free lunch: retrieval-augmented generation undermines fairness in llms, even for vigilant users. In: Findings of the Association for Computational Linguistics: EMNLP 2025. Association for Computational Linguistics; 2025:18145-18170. [doi: 10.18653\/v1\/2025.findings-emnlp.984]","DOI":"10.18653\/v1\/2025.findings-emnlp.984"},{"key":"R23","doi-asserted-by":"crossref","unstructured":"Zhang T Zhou Y Bollegala D . Evaluating the effect of retrieval augmentation on social biases. In: Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics. Association for Computational Linguistics; 2026:5004-5026. [doi: 10.18653\/v1\/2026.eacl-long.233]","DOI":"10.18653\/v1\/2026.eacl-long.233"},{"issue":"1","key":"R24","doi-asserted-by":"publisher","DOI":"10.1186\/s12939-025-02419-0","article-title":"Evaluating and addressing demographic disparities in medical large language models: a systematic review","volume":"24","author":"Omar","journal-title":"Int J Equity Health"},{"issue":"11","key":"R25","doi-asserted-by":"publisher","first-page":"7998","DOI":"10.1007\/s00330-022-08784-6","article-title":"Radiology artificial intelligence: a systematic review and evaluation of methods (RAISE)","volume":"32","author":"Kelly","journal-title":"Eur Radiol"},{"issue":"1","key":"R26","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-021-00438-z","article-title":"Diagnostic accuracy of deep learning in medical imaging: a systematic review and meta-analysis","volume":"4","author":"Aggarwal","journal-title":"NPJ Digit Med"},{"issue":"1","key":"R27","doi-asserted-by":"publisher","first-page":"63","DOI":"10.1148\/radiol.212966","article-title":"Deep learning algorithms to detect fractures: systematic review shows promising results but many limitations","volume":"304","author":"Cohen","journal-title":"Radiology"},{"issue":"6","key":"R28","doi-asserted-by":"publisher","first-page":"e271","DOI":"10.1016\/S2589-7500(19)30123-2","article-title":"A comparison of deep learning performance against health-care professionals in detecting diseases from medical imaging: a systematic review and meta-analysis","volume":"1","author":"Liu","journal-title":"Lancet Digit Health"},{"key":"R29","doi-asserted-by":"publisher","DOI":"10.1136\/bmj.m689","article-title":"Artificial intelligence versus clinicians: systematic review of design, reporting standards, and claims of deep learning studies","volume":"368","author":"Nagendran","journal-title":"BMJ"},{"issue":"2","key":"R30","doi-asserted-by":"publisher","DOI":"10.1161\/CIRCIMAGING.123.015495","article-title":"Machine learning and bias in medical imaging: opportunities and challenges","volume":"17","author":"Vrudhula","journal-title":"Circ Cardiovasc Imaging"},{"issue":"2","key":"R31","doi-asserted-by":"publisher","first-page":"75","DOI":"10.4274\/dir.2024.242854","article-title":"Bias in artificial intelligence for medical imaging: fundamentals, detection, avoidance, mitigation, challenges, ethics, and prospects","volume":"31","author":"Ko\u00e7ak","journal-title":"Diagn Interv Radiol"},{"issue":"1","key":"R32","doi-asserted-by":"publisher","DOI":"10.1186\/s12911-021-01533-7","article-title":"A systematic review of natural language processing applied to radiology reports","volume":"21","author":"Casey","journal-title":"BMC Med Inform Decis Mak"},{"issue":"1","key":"R33","doi-asserted-by":"publisher","DOI":"10.1038\/s43856-024-00717-2","article-title":"Current applications and challenges in large language models for patient care: a systematic review","volume":"5","author":"Busch","journal-title":"Commun Med (Lond)"},{"issue":"3","key":"R34","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"},{"key":"R35","unstructured":"JBI manual for evidence synthesis. Joanna Briggs Institute; 2024. URL: https:\/\/hotus.fi\/wp-content\/uploads\/2025\/10\/jbi-manual-for-evidence-synthesis-nov-2024.pdf [Accessed 08-08-2026]"},{"key":"R36","unstructured":"Vaswani A Shazeer N Parmar N et al. Attention is all you need. Presented at: NIPS\u201917: Proceedings of the 31st International Conference on Neural Information Processing Systems; Dec 4-9, 2017; Long Beach, California, USA. [doi: 10.5555\/3295222.3295349]"},{"issue":"1","key":"R37","doi-asserted-by":"publisher","DOI":"10.1186\/2046-4053-4-1","article-title":"Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement","volume":"4","author":"Moher","journal-title":"Syst Rev"},{"key":"R38","doi-asserted-by":"publisher","DOI":"10.1136\/bmj.i4919","article-title":"ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions","volume":"355","author":"Sterne","journal-title":"BMJ"},{"issue":"8","key":"R39","doi-asserted-by":"publisher","first-page":"529","DOI":"10.7326\/0003-4819-155-8-201110180-00009","article-title":"QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies","volume":"155","author":"Whiting","journal-title":"Ann Intern Med"},{"issue":"4\/5","key":"R40","doi-asserted-by":"publisher","first-page":"4","DOI":"10.1147\/JRD.2019.2942287","article-title":"AI Fairness 360: an extensible toolkit for detecting and mitigating algorithmic bias","volume":"63","author":"Bellamy","journal-title":"IBM J Res & Dev"},{"issue":"1","key":"R41","doi-asserted-by":"publisher","first-page":"51","DOI":"10.7326\/M18-1376","article-title":"PROBAST: a tool to assess the risk of bias and applicability of prediction model studies","volume":"170","author":"Wolff","journal-title":"Ann Intern Med"},{"issue":"4","key":"R42","doi-asserted-by":"publisher","DOI":"10.1002\/widm.1312","article-title":"Causability and explainability of artificial intelligence in medicine","volume":"9","author":"Holzinger","journal-title":"Wiley Interdiscip Rev Data Min Knowl Discov"},{"key":"R43","unstructured":"Lundberg SM Lee SI . A unified approach to interpreting model predictions. Presented at: NIPS\u201917: Proceedings of the 31st International Conference on Neural Information Processing Systems; Dec 4-9, 2017; Long Beach, California, USA. [doi: 10.5555\/3295222.3295230]"},{"key":"R44","doi-asserted-by":"crossref","unstructured":"Ribeiro MT Singh S Guestrin C . \u201cWhy should I trust you?\u201d: explaining the predictions of any classifier. Presented at: KDD \u201916: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining; Aug 13-17, 2016; San Francisco, California, USA. [doi: 10.1145\/2939672.2939778]","DOI":"10.1145\/2939672.2939778"},{"issue":"2","key":"R45","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3639372","article-title":"Explainability for large language models: a survey","volume":"15","author":"Zhao","journal-title":"ACM Trans Intell Syst Technol"},{"key":"R46","unstructured":"Kim B Wattenberg M Gilmer J et al. Interpretability beyond feature attribution: quantitative testing with concept activation vectors (TCAV). Presented at: 35th International Conference on Machine Learning (ICML 2018); Jul 10-15, 2018; Stockholm, Sweden. URL: https:\/\/proceedings.mlr.press\/v80\/kim18d\/kim18d.pdf [Accessed 08-08-2026]"},{"key":"R47","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"}],"container-title":["JMIR Research Protocols"],"original-title":[],"language":"en","deposited":{"date-parts":[[2026,9,10]],"date-time":"2026-09-10T21:13:51Z","timestamp":1789074831000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.researchprotocols.org\/2026\/1\/e91348"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,10]]},"references-count":47,"URL":"https:\/\/doi.org\/10.2196\/91348","relation":{},"ISSN":["1929-0748"],"issn-type":[{"value":"1929-0748","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,10]]},"article-number":"v15i11e91348"}}