{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,23]],"date-time":"2026-10-23T06:41:22Z","timestamp":1792737682515,"version":"4.5.0"},"reference-count":41,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T00:00:00Z","timestamp":1787702400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Med."],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>The management of pharmacotherapy in patients with multimorbidity and polypharmacy is a difficult task in routine clinical practice. Clinicians must consider diagnoses, prescribed drugs, contraindications, drug-drug interactions, renal and hepatic function, laboratory values, dose limits and individual risk factors. Clinical decision support systems in traditional settings are based on static rules, and hard to update. Standalone large language models are capable of processing clinical language but their outputs may be unsupported by evidence, incomplete or out of date. In this study, we developed and evaluated a hybrid multi-agent GraphRAG framework for personalised pharmacotherapy safety verification.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>We developed a system that integrates graph based retrieval with large language models. The framework was composed of three parts: (1) a schema for a pharmacotherapy knowledge graph that encoded indications, contraindications, interactions, laboratory thresholds, dose limitations and relevant patient conditions; (2) an Extractor\u2013Critic workflow that transformed unstructured medical text into graph elements with quality checks prior to their incorporation; and (3) a retrieval and re-ranking module that leveraged semantic search, graph traversal and safety-directed scoring. The system was evaluated on 12 benchmark cases extraction and 30 clinical safety questions answering. Comparison of conventional RAG and GraphRAG was done using linear mixed effects models, permutation testing, and paired case-level analysis.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      The multi-agent extraction workflow beat the zero-shot baseline. Strict F1 improved by 0.156 points (\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&amp;lt;\u20090.01), mainly due to higher ecall (+0.172,\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&amp;lt;\u20090.001). For clinical question answering, we found that GraphRAG outperformed conventional RAG in terms of medical accuracy (0.545 vs. 0.443, absolute difference of 0.102,\n                      <jats:italic>p<\/jats:italic>\n                      \u2009=\u20090.04). GraphRAG provided the better answer 66.7% of the time. The advantage was greatest where safety was dependent on contra-indications, laboratory cut-offs or patient specific factors. Smaller models performed as well as the larger models with the same GraphRAG configuration.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>The evaluation revealed that the integration of LLM-based extraction with a pharmacotherapy knowledge graph enhanced safety-oriented clinical decision support. GraphRAG provided more accurate and traceable answers than traditional RAG. Routine adoption will require validation on larger graphs, a wider range of datasets and real clinical populations, but this approach may be appropriate for cost sensitive and privacy conscious workflows.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fmed.2026.1898857","type":"journal-article","created":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T04:42:52Z","timestamp":1787719372000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["A multi-agent GraphRAG framework for pharmacotherapy safety verification in clinical decision support systems"],"prefix":"10.3389","volume":"13","author":[{"given":"Victor","family":"Ryzhenko","sequence":"first","affiliation":[{"name":"Department of Medical and Pharmaceutical Informatics and Advanced Technologies, Zaporizhzhia State Medical and Pharmaceutical University","place":["Zaporizhzhia, Ukraine"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bogdan","family":"Burlaka","sequence":"additional","affiliation":[{"name":"Department of Medicines Technology, Zaporizhzhia State Medical and Pharmaceutical University","place":["Zaporizhzhia, Ukraine"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Igor","family":"Belenichev","sequence":"additional","affiliation":[{"name":"Department of Pharmacology and Medical Formulation with Course of Normal Physiology, Zaporizhzhia State Medical and Pharmaceutical University","place":["Zaporizhzhia, Ukraine"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Kristina","family":"Burlaka","sequence":"additional","affiliation":[{"name":"Department of Clinical Laboratory Diagnostics, Zaporizhzhia State Medical and Pharmaceutical University","place":["Zaporizhzhia, Ukraine"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Olena","family":"Aliyeva","sequence":"additional","affiliation":[{"name":"Department of Histology, Cytology and Embryology, Zaporizhzhia State Medical and Pharmaceutical University","place":["Zaporizhzhia, Ukraine"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nina","family":"Bukhtiyarova","sequence":"additional","affiliation":[{"name":"Department of Clinical Laboratory Diagnostics, Zaporizhzhia State Medical and Pharmaceutical University","place":["Zaporizhzhia, Ukraine"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Iryna","family":"Halabitska","sequence":"additional","affiliation":[{"name":"Department of Therapy and Family Medicine, I. Horbachevsky Ternopil National Medical University","place":["Ternopil, Ukraine"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Pavlo","family":"Petakh","sequence":"additional","affiliation":[{"name":"Department of Biochemistry and Pharmacology, Uzhhorod National University","place":["Uzhhorod, Ukraine"]},{"name":"Kyiv School of Economics","place":["Kyiv, Ukraine"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Oleksandr","family":"Kamyshnyi","sequence":"additional","affiliation":[{"name":"Kyiv School of Economics","place":["Kyiv, Ukraine"]},{"name":"Department of Microbiology, Virology, and Immunology, I. Horbachevsky Ternopil National Medical University","place":["Ternopil, Ukraine"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,8,26]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"86","DOI":"10.1007\/s10462-025-11487-1","article-title":"Artificial intelligence-generated content (AIGC) in biomedical research, healthcare delivery, and clinical practices: technologies, applications, and regulatory considerations","volume":"59","author":"Ye","year":"2026","journal-title":"Artif Intell Rev"},{"key":"B2","doi-asserted-by":"publisher","first-page":"631","DOI":"10.3390\/bioengineering12060631","article-title":"Large language models in healthcare and medical applications: a review","volume":"12","author":"Maity","year":"2025","journal-title":"Bioengineering (Basel)"},{"key":"B3","doi-asserted-by":"publisher","first-page":"109713","DOI":"10.1016\/j.isci.2024.109713","article-title":"The application of large language models in medicine: a scoping review","volume":"27","author":"Meng","year":"2024","journal-title":"iScience"},{"key":"B4","doi-asserted-by":"publisher","first-page":"161","DOI":"10.1080\/10408363.2019.1678568","article-title":"Polypharmacy: a healthcare conundrum with a pharmacogenetic solution","volume":"57","author":"Sharp","year":"2020","journal-title":"Crit Rev Clin Lab Sci"},{"key":"B5","doi-asserted-by":"publisher","first-page":"e0253645","DOI":"10.1371\/journal.pone.0253645","article-title":"Analysis of interaction risks of patients with polypharmacy and the pharmacist interventions performed to solve them-A multicenter descriptive study according to medication reviews in Hungarian community pharmacies","volume":"16","author":"Szilvay","year":"2021","journal-title":"PLoS One"},{"key":"B6","doi-asserted-by":"publisher","first-page":"601","DOI":"10.1186\/s12877-022-03279-x","article-title":"Prevalence and factors associated with polypharmacy: a systematic review and meta-analysis","volume":"22","author":"Delara","year":"2022","journal-title":"BMC Geriatr"},{"key":"B7","doi-asserted-by":"publisher","first-page":"3235","DOI":"10.2147\/DDDT.S257494","article-title":"Potential impact of the multi-target drug approach in the treatment of some Complex diseases","volume":"14","author":"Makhoba","year":"2020","journal-title":"Drug Des Devel Ther"},{"key":"B8","doi-asserted-by":"publisher","first-page":"395","DOI":"10.3390\/ph17030395","article-title":"Digital technology applications in the management of adverse drug reactions: bibliometric analysis","volume":"17","author":"Litvinova","year":"2024","journal-title":"Pharmaceuticals (Basel)"},{"key":"B9","doi-asserted-by":"publisher","first-page":"47","DOI":"10.1186\/s41235-023-00497-8","article-title":"Cognitive perspectives on maintaining physicians\u2019 medical expertise: iI. Acquiring, maintaining, and updating cognitive skills","volume":"8","author":"Caddick","year":"2023","journal-title":"Cogn Res Princ Implic"},{"key":"B10","doi-asserted-by":"publisher","first-page":"23333928241234863","DOI":"10.1177\/23333928241234863","article-title":"Artificial intelligence and decision-making in healthcare: a thematic analysis of a systematic review of reviews","volume":"11","author":"Khosravi","year":"2024","journal-title":"Health Serv Res Manag Epidemiol"},{"key":"B11","doi-asserted-by":"publisher","first-page":"100117","DOI":"10.1016\/j.sintl.2021.100117","article-title":"Telemedicine for healthcare: capabilities, features, barriers, and applications","volume":"2","author":"Haleem","year":"2021","journal-title":"Sens Int"},{"key":"B12","doi-asserted-by":"publisher","first-page":"96","DOI":"10.3390\/informatics11040096","article-title":"Transforming service quality in healthcare: a comprehensive review of healthcare 4.0 and its impact on healthcare service quality","volume":"11","author":"Al-Assaf","year":"2024","journal-title":"Informatics"},{"key":"B13","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","year":"2019","journal-title":"Nat Med"},{"key":"B14","doi-asserted-by":"publisher","first-page":"17","DOI":"10.1038\/s41746-020-0221-y","article-title":"An overview of clinical decision support systems: benefits, risks, and strategies for success","volume":"3","author":"Sutton","year":"2020","journal-title":"NPJ Digit Med"},{"key":"B15","doi-asserted-by":"publisher","first-page":"2199","DOI":"10.1001\/jama.2018.17163","article-title":"Clinical decision support in the era of artificial intelligence","volume":"320","author":"Shortliffe","year":"2018","journal-title":"JAMA"},{"key":"B16","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","year":"2024","journal-title":"Nat Med"},{"key":"B17","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s12553-023-00806-7","article-title":"The underuse of AI in the health sector: opportunity costs, success stories, risks and recommendations","volume":"14","author":"Pagallo","year":"2024","journal-title":"Health Technol (Berl)"},{"key":"B18","doi-asserted-by":"publisher","first-page":"S23","DOI":"10.15265\/IYS-2016-s004","article-title":"Imaging informatics: 25 years of progress","volume":"25","author":"Agrawal","year":"2016","journal-title":"Yearb Med Inform"},{"key":"B19","doi-asserted-by":"publisher","first-page":"105246","DOI":"10.1016\/j.ijmedinf.2023.105246","article-title":"Machine learning models to detect and predict patient safety events using electronic health records: a systematic review","volume":"180","author":"Deimazar","year":"2023","journal-title":"Int J Med Inform"},{"key":"B20","doi-asserted-by":"publisher","first-page":"321","DOI":"10.1007\/s40264-024-01505-6","article-title":"Leveraging natural language processing and machine learning methods for adverse drug event detection in electronic health\/medical records: a scoping review","volume":"48","author":"Golder","year":"2025","journal-title":"Drug Saf"},{"key":"B21","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","year":"2020","journal-title":"J Am Med Inform Assoc"},{"key":"B22","doi-asserted-by":"publisher","first-page":"e52897","DOI":"10.2196\/52897","article-title":"Multifaceted natural language processing task\u2013based evaluation of bidirectional encoder representations from transformers models for bilingual (Korean and English) clinical notes: algorithm development and validation","volume":"12","author":"Kim","year":"2024","journal-title":"JMIR Med Inform"},{"key":"B23","doi-asserted-by":"crossref","DOI":"10.1145\/3688671.3688756","article-title":"Prediction of drug-drug interactions with zero-shot learning","author":"Astras","year":"2024"},{"key":"B24","doi-asserted-by":"publisher","first-page":"bbae354","DOI":"10.1093\/bib\/bbae354","article-title":"Harnessing large language models\u2019 zero-shot and few-shot learning capabilities for regulatory research","volume":"25","author":"Meshkin","year":"2024","journal-title":"Brief Bioinform"},{"key":"B25","doi-asserted-by":"publisher","first-page":"45","DOI":"10.1038\/s41746-024-01377-1","article-title":"Clinical entity augmented retrieval for clinical information extraction","volume":"8","author":"Lopez","year":"2025","journal-title":"NPJ Digit Med"},{"key":"B26","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","year":"2023","journal-title":"Nat Med"},{"key":"B27","doi-asserted-by":"publisher","first-page":"10.1056\/aioa2300068","DOI":"10.1056\/aioa2300068","article-title":"Almanac\u2014retrieval-augmented language models for clinical medicine","volume":"1","author":"Zakka","year":"2024","journal-title":"NEJM AI"},{"key":"B28","author":"Gao","year":"2024"},{"key":"B29","author":"Lewis","year":"2020"},{"key":"B30","author":"Menick","year":"2022"},{"key":"B31","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","year":"2023","journal-title":"Nature"},{"key":"B32","author":"Wei","year":"2023"},{"key":"B33","author":"Taylor","year":"2022"},{"key":"B34","author":"Frieder","year":"2023"},{"key":"B35","author":"Noori","year":"2025"},{"key":"B36","author":"Jing","year":"2026"},{"key":"B37","author":"Huang","year":"2024"},{"key":"B38","doi-asserted-by":"publisher","first-page":"eady2539","DOI":"10.1126\/sciadv.ady2539","article-title":"Multidimensional mapping of stimulation-responsive regulatory elements and candidate causal variants in T cell activation","volume":"12","author":"Wang","year":"2026","journal-title":"Sci Adv"},{"key":"B39","doi-asserted-by":"publisher","first-page":"26","DOI":"10.1186\/s40537-025-01330-3","article-title":"Meta-analysis of large language models: benchmarking DeepSeek-R1 against ChatGPT, gemini, qwen, and LLaMA","volume":"13","author":"Awan","year":"2026","journal-title":"J Big Data"},{"key":"B40","author":"Yao","year":"2022"},{"key":"B41","author":"Edge","year":"2024"}],"container-title":["Frontiers in Medicine"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1898857\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T04:43:02Z","timestamp":1787719382000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1898857\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,26]]},"references-count":41,"alternative-id":["10.3389\/fmed.2026.1898857"],"URL":"https:\/\/doi.org\/10.3389\/fmed.2026.1898857","relation":{},"ISSN":["2296-858X"],"issn-type":[{"value":"2296-858X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,26]]},"article-number":"1898857"}}