{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T18:40:43Z","timestamp":1788288043573,"version":"build-2803163510"},"reference-count":50,"publisher":"Wiley","issue":"10","license":[{"start":{"date-parts":[[2025,9,17]],"date-time":"2025-09-17T00:00:00Z","timestamp":1758067200000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"publisher","award":["DK125718"],"award-info":[{"award-number":["DK125718"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Liver International"],"published-print":{"date-parts":[[2025,10]]},"abstract":"<jats:title>ABSTRACT<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Background and Aims<\/jats:title>\n                    <jats:p>Advances in artificial intelligence, particularly large language models (LLMs), hold promise for transforming chronic disease management such as Hepatitis C Virus (HCV) infection. This study evaluates the impact of retrieval\u2010augmented generation (RAG) and supervised fine\u2010tuning (SFT) on both open\u2010ended question answering (accuracy and clarity) and on LLM\u2010recommended treatment regimens for clinical scenarios.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>We employed OpenAI's GPT\u20104 Turbo in four configurations\u2014baseline, RAG\u2010Top1, RAG\u2010Top 10 and SFT\u2014using the 2020 EASL HCV guidelines as external knowledge or fine\u2010tuning data. For the question set, guidelines were segmented at the paragraph level and encoded into 3072\u2010dimensional embeddings. Fifteen questions covering general, patient and physician perspectives were scored on a 10\u2010point accuracy scale and binary accuracy\/clarity by four experts. Separately, we created 25 simulated clinical scenarios; a consensus of four hepatologists defined the gold\u2010standard DAA regimens. Model performance on these cases was measured by two metrics: \u2018partial accuracy\u2019 (\u2265\u2009one correct DAA without errors) and \u2018complete accuracy\u2019 (all correct DAAs without errors).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      On open\u2010ended questions, RAG\u2010Top10 outperformed baseline in accuracy (91.7% vs. 36.6%;\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&lt;\u20090.001) and clarity (91.7% vs. 46.6%;\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&lt;\u20090.001). RAG\u2010Top1 achieved 81.7% accuracy and 86.6% clarity (both\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&lt;\u20090.001), while SFT reached 71.7% accuracy and 88.3% clarity (\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&lt;\u20090.001). Similarly, RAG\u2010Top10 achieved the highest performance in prescribing the correct DAA regimen according to expert consensus in 76% of cases (vs. 24% for baseline model,\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&lt;\u20090.001).\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>Both RAG\u2010Top10 and SFT markedly enhance LLM performance in guideline\u2010driven HCV management\u2014improving not only response accuracy and clarity but also DAA selection in clinical scenarios. RAG\u2010Top10's broader context retrieval confers the greatest gains, while SFT underscores the value of domain\u2010specific alignment. Rigorous, expert\u2010informed evaluation frameworks are essential for the safe integration of LLMs into clinical practice.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1111\/liv.70349","type":"journal-article","created":{"date-parts":[[2025,9,17]],"date-time":"2025-09-17T12:55:26Z","timestamp":1758113726000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":5,"title":["From Guidelines to Real\u2010Time Conversation: Expert\u2010Validated Retrieval\u2010Augmented and Fine\u2010Tuned\n                    <scp>GPT<\/scp>\n                    \u20104 for Hepatitis C Management"],"prefix":"10.1111","volume":"45","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-9910-3514","authenticated-orcid":false,"given":"Mauro","family":"Giuffr\u00e8","sequence":"first","affiliation":[{"name":"Department of Internal Medicine (Section of Digestive Diseases), Yale School of Medicine Yale University  New Haven Connecticut USA"},{"name":"Department of Medicine, Surgery and Health Sciences University of Trieste  Trieste Italy"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nicola","family":"Pugliese","sequence":"additional","affiliation":[{"name":"Department of Biomedical Sciences Humanitas University  Milano Italy"},{"name":"Division of Internal Medicine and Hepatology, Department of Gastroenterology IRCCS Humanitas Research Hospital  Milano Italy"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Simone","family":"Kresevic","sequence":"additional","affiliation":[{"name":"Department of Engineering and Architecture University of Trieste  Trieste Italy"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Milos","family":"Ajcevic","sequence":"additional","affiliation":[{"name":"Department of Engineering and Architecture University of Trieste  Trieste Italy"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Francesco","family":"Negro","sequence":"additional","affiliation":[{"name":"Division of Gastroenterology University Hospitals of Geneva  Geneva Switzerland"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3278-7138","authenticated-orcid":false,"given":"Massimo","family":"Puoti","sequence":"additional","affiliation":[{"name":"Division of Infectious Diseases ASST GOM Niguarda  Milan Italy"},{"name":"Department of Medicine University of Milan Bicocca  Milan Italy"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xavier","family":"Forns","sequence":"additional","affiliation":[{"name":"Liver Unit, Hospital Cl\u00ednic University of Barcelona, IDIBAPS  Barcelona Spain"},{"name":"Consorcio de Investigaci\u00f3n Biom\u00e9dica en Red de Enfermedades Hep\u00e1ticas y Digestivas, ISCIII  Madrid Spain"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0745-7559","authenticated-orcid":false,"given":"Jean\u2010Michel","family":"Pawlotsky","sequence":"additional","affiliation":[{"name":"Universit\u00e9 Paris Est Cr\u00e9teil, INSERM, IMRB  Cr\u00e9teil France"},{"name":"Inserm, U955, Team18  Cr\u00e9teil France"},{"name":"Department of Digestive and Hepatobiliary Surgery Assistance Publique\u2010H\u00f4pitaux de Paris, Paul Brousse University Hospital  Paris France"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Dennis L.","family":"Shung","sequence":"additional","affiliation":[{"name":"Department of Internal Medicine (Section of Digestive Diseases), Yale School of Medicine Yale University  New Haven Connecticut USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0941-3226","authenticated-orcid":false,"given":"Alessio","family":"Aghemo","sequence":"additional","affiliation":[{"name":"Department of Biomedical Sciences Humanitas University  Milano Italy"},{"name":"Division of Internal Medicine and Hepatology, Department of Gastroenterology IRCCS Humanitas Research Hospital  Milano Italy"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2025,9,17]]},"reference":[{"key":"e_1_2_11_2_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41746\u2010023\u201000958\u2010w"},{"key":"e_1_2_11_3_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41591\u2010023\u201002448\u20108"},{"key":"e_1_2_11_4_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41746\u2010023\u201000873\u20100"},{"key":"e_1_2_11_5_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41586\u2010023\u201006291\u20102"},{"key":"e_1_2_11_6_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41591\u2010023\u201002700\u20101"},{"key":"e_1_2_11_7_1","doi-asserted-by":"publisher","DOI":"10.1093\/cid\/ciad319"},{"key":"e_1_2_11_8_1","doi-asserted-by":"publisher","DOI":"10.1001\/jamanetworkopen.2022.45424"},{"key":"e_1_2_11_9_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.jhep.2020.08.018"},{"issue":"18","key":"e_1_2_11_10_1","first-page":"357","article-title":"Vital Signs: Evaluation of Hepatitis C Virus Infection Testing and Reporting \u2010 Eight U.S. Sites, 2005\u20132011","volume":"62","author":"Centers for Disease Control and Prevention (CDC)","year":"2013","journal-title":"MMWR. Morbidity and Mortality Weekly Report"},{"key":"e_1_2_11_11_1","doi-asserted-by":"publisher","DOI":"10.1097\/QMH.0000000000000007"},{"key":"e_1_2_11_12_1","doi-asserted-by":"publisher","DOI":"10.1007\/s00404\u2010023\u201007146\u2010x"},{"key":"e_1_2_11_13_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41467\u2010024\u201045563\u2010x"},{"key":"e_1_2_11_14_1","doi-asserted-by":"publisher","DOI":"10.1111\/apt.18058"},{"key":"e_1_2_11_15_1","doi-asserted-by":"publisher","DOI":"10.1053\/j.gastro.2025.03.038"},{"key":"e_1_2_11_16_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.cgh.2024.01.024"},{"key":"e_1_2_11_17_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.cgh.2023.09.035"},{"key":"e_1_2_11_18_1","unstructured":"P.Lewis E.Perez A.Piktus et\u00a0al. \u201cRetrieval\u2010Augmented Generation for Knowledge\u2010Intensive NLP Tasks \u201d arXiv (2005)."},{"key":"e_1_2_11_19_1","unstructured":"M. R.Douglas \u201cLarge Language Models \u201d arXiv (2023)."},{"key":"e_1_2_11_20_1","doi-asserted-by":"crossref","unstructured":"K.Lv Y.Yang T.Liu Q.Gao Q.Guo andX.Qiu \u201cFull Parameter Fine\u2010Tuning for Large Language Models With Limited Resources \u201d arXiv (2023).","DOI":"10.18653\/v1\/2024.acl-long.445"},{"key":"e_1_2_11_21_1","unstructured":"D. M.Ziegler N.Stiennon J.Wu et\u00a0al. \u201cFine\u2010Tuning Language Models From Human Preferences \u201d arXiv (2019)."},{"key":"e_1_2_11_22_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41746\u2010025\u201001589\u2010z"},{"key":"e_1_2_11_23_1","doi-asserted-by":"publisher","DOI":"10.1111\/liv.15974"},{"key":"e_1_2_11_24_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41746\u2010025\u201001896\u20105"},{"key":"e_1_2_11_25_1","unstructured":"P.Christiano J.Leike T. B.Brown M.Martic S.Legg andD.Amodei \u201cDeep Reinforcement Learning From Human Preferences \u201d arXiv (2017)."},{"key":"e_1_2_11_26_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.jhepr.2025.101523"},{"key":"e_1_2_11_27_1","doi-asserted-by":"publisher","DOI":"10.1111\/jgh.16375"},{"key":"e_1_2_11_28_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41746\u2010024\u201001091\u2010y"},{"key":"e_1_2_11_29_1","doi-asserted-by":"publisher","DOI":"10.1016\/S0001\u20106918(99)00050\u20105"},{"key":"e_1_2_11_30_1","doi-asserted-by":"publisher","DOI":"10.1177\/0049124197025003003"},{"key":"e_1_2_11_31_1","doi-asserted-by":"crossref","unstructured":"A.SimhiandS.Markovitch \u201cInterpreting Embedding Spaces by Conceptualization \u201d arXiv (2022).","DOI":"10.18653\/v1\/2023.emnlp-main.106"},{"key":"e_1_2_11_32_1","unstructured":"OpenAI \u201cEmbeddings \u201d accessed March 10 2024 https:\/\/platform.openai.com\/docs\/guides\/embeddings."},{"key":"e_1_2_11_33_1","doi-asserted-by":"crossref","unstructured":"S.Wang S.Li T.Sun et\u00a0al. \u201cLLM Can Achieve Self\u2010Regulation via Hyperparameter Aware Generation \u201d arXiv (2024).","DOI":"10.18653\/v1\/2024.findings-acl.396"},{"key":"e_1_2_11_34_1","unstructured":"B.Atil A.Chittams L.Fu F.Ture L.Xu andB.Baldwin \u201cLLM Stability: A Detailed Analysis With Some Surprises \u201d arXiv (2024)."},{"key":"e_1_2_11_35_1","unstructured":"A.Radford J.Wu R.Child D.Luan D.Amodei andI.Sutskever \u201cLanguage Models Are Unsupervised Multitask Learners \u201dhttps:\/\/github.com\/codelucas\/newspaper."},{"key":"e_1_2_11_36_1","doi-asserted-by":"publisher","DOI":"10.14309\/ajg.0000000000003439"},{"key":"e_1_2_11_37_1","unstructured":"A. J.Yepes Y.You J.Milczek S.Laverde andR.Li \u201cFinancial Report Chunking for Effective Retrieval Augmented Generation \u201d arXiv (2024)."},{"key":"e_1_2_11_38_1","unstructured":"M.Giuffr\u00e8andS.Kresevic \u201cHCV\u2010GPT GitHub Repository by LIVER\u2010MAINDS Laboratory \u201d accessed July 8 2025 https:\/\/github.com\/liver\u2010mainds\/hcv\u2010gpt."},{"key":"e_1_2_11_39_1","unstructured":"OpenAI \u201cFine\u2010Tuning \u201d accessed May 03 2024 https:\/\/platform.openai.com\/docs\/guides\/fine\u2010tuning."},{"key":"e_1_2_11_40_1","unstructured":"L.Ouyang J.Wu X.Jiang et\u00a0al. \u201cTraining Language Models to Follow Instructions With Human Feedback \u201d arXiv (2017)."},{"key":"e_1_2_11_41_1","doi-asserted-by":"publisher","DOI":"10.1037\/0033\u20102909.86.2.420"},{"key":"e_1_2_11_42_1","doi-asserted-by":"publisher","DOI":"10.1037\/1040\u20103590.6.4.284"},{"key":"e_1_2_11_43_1","doi-asserted-by":"publisher","DOI":"10.1037\/h0031619"},{"key":"e_1_2_11_44_1","doi-asserted-by":"crossref","unstructured":"G.IzacardandE.Grave \u201cLeveraging Passage Retrieval With Generative Models for Open Domain Question Answering \u201d arXiv (2020).","DOI":"10.18653\/v1\/2021.eacl-main.74"},{"key":"e_1_2_11_45_1","unstructured":"P.Lewis E.Perez A.Piktus et\u00a0al. \u201cRetrieval\u2010Augmented Generation for Knowledge\u2010Intensive NLP Tasks \u201d arXiv (2020)."},{"key":"e_1_2_11_46_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.cgh.2023.11.008"},{"key":"e_1_2_11_47_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.gie.2023.06.025"},{"key":"e_1_2_11_48_1","doi-asserted-by":"publisher","DOI":"10.1053\/j.gastro.2023.08.033"},{"key":"e_1_2_11_49_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41591\u2010024\u201002855\u20105"},{"key":"e_1_2_11_50_1","doi-asserted-by":"publisher","DOI":"10.1016\/S2589\u20107500(23)00202\u20109"},{"key":"e_1_2_11_51_1","doi-asserted-by":"publisher","DOI":"10.1097\/HEP.0000000000000992"}],"container-title":["Liver International"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/liv.70349","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,17]],"date-time":"2025-12-17T16:41:39Z","timestamp":1765989699000},"score":1,"resource":{"primary":{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/10.1111\/liv.70349"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,9,17]]},"references-count":50,"journal-issue":{"issue":"10","published-print":{"date-parts":[[2025,10]]}},"alternative-id":["10.1111\/liv.70349"],"URL":"https:\/\/doi.org\/10.1111\/liv.70349","archive":["Portico"],"relation":{},"ISSN":["1478-3223","1478-3231"],"issn-type":[{"value":"1478-3223","type":"print"},{"value":"1478-3231","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,9,17]]},"assertion":[{"value":"2025-01-10","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-09-03","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-09-17","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e70349"}}