{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,29]],"date-time":"2026-09-29T20:05:02Z","timestamp":1790712302725,"version":"4.1.0"},"reference-count":53,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["JMIR Med Inform"],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Misspellings in medication names can compromise patient safety, reduce data utility, and impede large-scale data initiatives that integrate medication information from electronic health records (EHRs). Existing methods for detecting misspelled medical terms are mostly dictionary-based and can lead to high false-positive rates when correctly spelled but previously unseen (out-of-vocabulary) terms are encountered.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>We aimed to develop and validate domain-specific, transformer-based language models for detecting misspelled drug names, with an emphasis on performance for unseen terms.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>Using RxNorm as a standardized drug vocabulary, we created an RxNorm-augmented training corpus and developed two BERT (Bidirectional Encoder Representations from Transformers)\u2013based models\u2014BERTDrug and CharBERTDrug\u2014for misspelling detection. Specifically, we randomly split 69,824 RxNorm drug names into training, development, and test sets (3:1:1) and generated k misspellings per name using text-perturbation techniques (k optimized for training; fixed at 1 for development and test sets). The models were fine-tuned on the training and development sets and evaluated using the RxNorm test set and 3586 drug names from the Long-Term Care Data Cooperative (LTCDC) database (external validation). The RxNorm test set and out-of-vocabulary LTCDC dataset (1922 terms), neither overlapping with the RxNorm training data, were used to evaluate performance on unseen terms. SpellChecker served as a dictionary-based baseline, while fastTextML and BioWordVecML, which used different subword embeddings as inputs for machine learning, served as additional baselines. Additionally, we compared model performance with GPT-4o, a generative large language model (LLM), using 2200 randomly sampled test terms.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>On the RxNorm test set, BERTDrug and CharBERTDrug outperformed the baseline models across most metrics. BERTDrug achieved the best overall performance (F1-score=0.859; area under the receiver operating characteristic curve [ROC-AUC]=0.947), followed by CharBERTDrug (F1-score=0.833; ROC-AUC=0.906). Both models also outperformed the baseline models on the out-of-vocabulary LTCDC dataset across most metrics, with CharBERTDrug performing best (F1-score=0.696; ROC-AUC=0.788), followed by BERTDrug (F1-score=0.669; ROC-AUC=0.786). In the secondary analysis, both models exceeded GPT-4o on most metrics (except Recall) for 2000 RxNorm terms. BERTDrug performed best (ROC-AUC=0.951; F1-score=0.855), followed by CharBERTDrug (ROC-AUC=0.911; F1-score=0.831) and GPT-4o (ROC-AUC=0.856; F1-score=0.721). In contrast, among 200 randomly selected LTCDC terms, GPT-4o performed best on most metrics except precision and specificity.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>Domain-specific language models improved detection of misspellings in out-of-vocabulary drug names and outperformed baseline models in both internal and external evaluations. The comparison with a generative LLM suggests that domain shift may substantially reduce the advantages conferred by domain-specific training. With further fine-tuning on diverse data that capture the terminology, formatting conventions, and spelling patterns encountered across real-world clinical settings, these models could be adapted for use in other clinical databases and EHR systems to improve medication data quality for research and to support future safety-focused applications.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.2196\/91151","type":"journal-article","created":{"date-parts":[[2026,9,29]],"date-time":"2026-09-29T19:15:06Z","timestamp":1790709306000},"page":"e91151","source":"Crossref","is-referenced-by-count":0,"title":["Detecting Misspelled Drug Names Using Transformer-Based Language Models: Model Development and External Validation"],"prefix":"10.2196","volume":"14","author":[{"ORCID":"https:\/\/orcid.org\/0009-0006-1617-7267","authenticated-orcid":false,"given":"Jiayu","family":"Lu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5056-0431","authenticated-orcid":false,"given":"Kevin W","family":"McConeghy","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1673-4570","authenticated-orcid":false,"given":"Andrew R","family":"Zullo","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7259-4301","authenticated-orcid":false,"given":"Jinying","family":"Chen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,9,29]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijmedinf.2010.09.005"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1055\/s-0041-1731745"},{"key":"ref3","unstructured":"Medication safety for look-alike, sound-alike medicinesWorld Health Organization202310202026-08-20https:\/\/www.who.int\/publications\/i\/item\/9789240058897"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1016\/j.pmedr.2024.102765"},{"key":"ref5","first-page":"1079","volume":"2012","author":"Zhou, L","year":"2012","journal-title":"AMIA Annu Symp Proc"},{"key":"ref6","first-page":"751","volume":"2007","author":"Turchin, A","year":"2007","journal-title":"AMIA Annu Symp Proc"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1016\/j.jbi.2015.04.008"},{"key":"ref8","doi-asserted-by":"crossref","first-page":"333","DOI":"10.5220\/0005911503330338","author":"Hussain, F","year":"2016","journal-title":"Proceedings of the 18th International Conference on Enterprise Information Systems"},{"key":"ref9","unstructured":"Medication errors linked to drug name confusionPatient Safety Authority20042026-08-20https:\/\/patientsafety.pa.gov\/ADVISORIES\/Pages\/200412_07.aspx"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1136\/amiajnl-2012-001241"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1016\/j.clinthera.2015.03.003"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1177\/1460458220941750"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1136\/amiajnl-2013-001745"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1136\/bmj.i4854"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1186\/1472-6947-7-3"},{"key":"ref16","first-page":"90","volume":"2014","author":"Pimpalkhute, P","year":"2014","journal-title":"AMIA Jt Summits Transl Sci Proc"},{"key":"ref17","first-page":"136","volume":"247","author":"Jiang, K","year":"2018","journal-title":"Stud Health Technol Inform"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1016\/j.jclinepi.2020.09.036"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1056\/NEJMsr1809937"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1016\/j.jamda.2022.09.006"},{"key":"ref21","unstructured":"PatrickJSabbaghMJainSZhengHSpelling correction in clinical notes with emphasis on first suggestion accuracy2010Proceedings of 2nd Workshop on Building and Evaluating Resources for Biomedical Text Mining (BioTxtM)2010Valletta, Malta"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1016\/j.jocs.2024.102393"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.3390\/s21092893"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1145\/146370.146380"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1109\/MITP.2005.122"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1055\/s-0038-1667077"},{"key":"ref27","doi-asserted-by":"crossref","first-page":"2836","DOI":"10.18653\/v1\/2021.emnlp-main.225","author":"Zhuang, S","year":"2021","journal-title":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing"},{"key":"ref28","first-page":"6903","author":"El Boukkouri, H","year":"2020","journal-title":"Proceedings of the 28th International Conference on Computational Linguistics"},{"key":"ref29","first-page":"4171","author":"Devlin, J","year":"2019","journal-title":"Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers)"},{"key":"ref30","author":"Aftan, S","year":"2023","journal-title":"2023 20th Learning and Technology Conference (L&T)"},{"key":"ref31","first-page":"6000","author":"Vaswani, A","year":"2017","journal-title":"Proceedings of the 31st International Conference on Neural Information Processing Systems"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1038\/sdata.2016.35"},{"key":"ref33","unstructured":"PMC Open Access SubsetNational Library of Medicine20232026-08-19https:\/\/pmc.ncbi.nlm.nih.gov\/tools\/openftlist\/"},{"key":"ref34","first-page":"2377","author":"Srivastava, R","year":"2015","journal-title":"Proceedings of the 29th International Conference on Neural Information Processing Systems"},{"key":"ref35","unstructured":"KrallingerMRabalOAkhondiSP\u00e9rezMSantamar\u00edaJRodr\u00edguezGOverview of the BioCreative VI chemical-protein interaction track2017Proceedings of the Sixth BioCreative Challenge Evaluation WorkshopOctober 18, 2017Bethesda, MA141146"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1016\/j.jbi.2013.07.011"},{"key":"ref37","author":"L\u00f3pez-Hern\u00e1ndez, J","year":"2019","journal-title":"Technologies and Innovation"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.3390\/electronics9101670"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1162\/tacl_a_00051"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1038\/s41597-019-0055-0"},{"key":"ref41","unstructured":"GitHub2026-08-19https:\/\/github.com\/ncbi-nlp\/BioSentVec#text-corpora"},{"key":"ref42","doi-asserted-by":"crossref","first-page":"882","DOI":"10.18653\/v1\/2020.acl-main.82","author":"Zhang, S","year":"2020","journal-title":"Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics"},{"key":"ref43","doi-asserted-by":"crossref","first-page":"146","DOI":"10.18653\/v1\/P18-3021","author":"Etoori, P","year":"2018","journal-title":"Proceedings of ACL 2018, Student Research Workshop"},{"key":"ref44","first-page":"158","author":"Jayanthi, SM","year":"2020","journal-title":"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1709.06429"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-023-47295-2"},{"key":"ref47","first-page":"85","author":"Klemen, M","year":"2024","journal-title":"International Conference on Text, Speech, and Dialogue. TSD 2024. Lecture Notes in Computer Science"},{"key":"ref48","unstructured":"Ul\u010darMRobnik \u0160ikonjaMSloBERTa: Slovene monolingual large pretrained masked language model2021Proceedings of Data Mining and Data WarehousingOct 4, 2021Ljubljana1720"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1016\/S2589-7500(23)00254-6"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-025-03727-2"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1093\/jamiaopen\/ooae154"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-024-01366-4"},{"key":"ref53","unstructured":"DatasetGitHub2026-09-08https:\/\/github.com\/jchen-BUlab\/NLP_misspelling_detect"}],"container-title":["JMIR Medical Informatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e91151","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,29]],"date-time":"2026-09-29T19:15:10Z","timestamp":1790709310000},"score":1,"resource":{"primary":{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e91151"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,29]]},"references-count":53,"URL":"https:\/\/doi.org\/10.2196\/91151","relation":{"has-preprint":[{"id-type":"doi","id":"10.2196\/preprints.91151","asserted-by":"object"}]},"ISSN":["2291-9694"],"issn-type":[{"value":"2291-9694","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,29]]}}}