{"doi":"10.1093/jamia/ocaf193","title":"Automated detection of stigmatizing language in Electronic Health Records (EHRs) using a multi-stage transfer learning approach","abstract":"<jats:title>Abstract</jats:title>\n                  <jats:sec>\n                    <jats:title>Objective</jats:title>\n                    <jats:p>Stigmatizing language (SL) in Electronic Health Records (EHRs) can perpetuate biases and negatively impact patient care. This study introduces a novel method for automatically detecting such language to improve healthcare documentation practices.</jats:p>\n                  </jats:sec>\n                  <jats:sec>\n                    <jats:title>Materials and Methods</jats:title>\n                    <jats:p>We developed a multi-stage transfer learning framework integrating semantic, syntactic, and task adaptation using three datasets: hate speech, clinical phenotypes, and stigmatizing language. Experiments were conducted on stigmatizing language dataset which consists of 4,129 de-identified EHR notes (72.7% stigmatizing, 27.3% non-stigmatizing), split 80/20 for training and testing. Longformer, BERT, and ClinicalBERT models were evaluated, and model performance was assessed on 35 randomized subsets of the test set (each comprising 70% of test data). The Wilcoxon-Mann-Whitney test was used to evaluate statistical significance, with Bonferroni correction applied to control for multiple hypothesis testing. Baseline models included zero-shot and few-shot GPT-4o, Support Vector Machine, Random Forest, Logistic Regression, and Multinomial Naive Bayes.</jats:p>\n                  </jats:sec>\n                  <jats:sec>\n                    <jats:title>Results</jats:title>\n                    <jats:p>The proposed framework achieved the highest accuracy, with fully adapted Longformer reaching 89.83%. Performance improvements remained statistically significant after Bonferroni correction compared to all baselines (p &amp;lt; .05). The framework demonstrated robust gains across different stigmatizing language types.</jats:p>\n                  </jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion</jats:title>\n                    <jats:p>This study underscores the value of domain-adaptive NLP for detecting stigmatizing language in EHRs. The multi-stage transfer learning framework effectively captures subtle biases often missed by conventional models, enabling more objective and respectful clinical documentation.</jats:p>\n                  </jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion</jats:title>\n                    <jats:p>This framework offers a statistically validated, high-performing framework for detecting stigmatizing language in EHRs, supporting responsible AI and promoting equity in clinical care.</jats:p>\n                  </jats:sec>","journal":"Journal of the American Medical Informatics Association","year":2026,"id":623791,"datarank":0.37334651683768527,"base_score":1.3862943611198906,"endowment":1.3862943611198906,"self_citation_contribution":0.20794415416798362,"citation_network_contribution":0.16540236266970162,"self_endowment_contribution":0.20794415416798362,"citer_contribution":0.16540236266970162,"corpus_percentile":null,"corpus_rank":null,"citation_count":3,"citer_count":1,"citers_with_citation_signal":1,"citers_with_endowment":1,"datacite_reuse_total":0,"is_dataset":false,"is_dataset_confidence":null,"is_data_producer":false,"deposit_databanks":null,"is_oa":false,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":null,"fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":1612301,"name":"A M Muntasir Rahman","orcid":null,"position":1,"is_corresponding":false},{"id":1612302,"name":"Charles R Senteio","orcid":"0000-0002-0254-3127","position":2,"is_corresponding":false},{"id":959554,"name":"Vivek K. Singh","orcid":"0000-0002-8194-2336","position":3,"is_corresponding":false},{"id":1612300,"name":"Liyang Xue","orcid":"0000-0003-2908-9072","position":0,"is_corresponding":false}],"reference_count":0,"raw_metadata":{"has_enrichment":true,"resolved":true,"title":"Automated detection of stigmatizing language in Electronic Health Records (EHRs) using a multi-stage transfer learning approach","abstract":"<jats:title>Abstract</jats:title>\n                  <jats:sec>\n                    <jats:title>Objective</jats:title>\n                    <jats:p>Stigmatizing language (SL) in Electronic Health Records (EHRs) can perpetuate biases and negatively impact patient care. This study introduces a novel method for automatically detecting such language to improve healthcare documentation practices.</jats:p>\n                  </jats:sec>\n                  <jats:sec>\n                    <jats:title>Materials and Methods</jats:title>\n                    <jats:p>We developed a multi-stage transfer learning framework integrating semantic, syntactic, and task adaptation using three datasets: hate speech, clinical phenotypes, and stigmatizing language. Experiments were conducted on stigmatizing language dataset which consists of 4,129 de-identified EHR notes (72.7% stigmatizing, 27.3% non-stigmatizing), split 80/20 for training and testing. Longformer, BERT, and ClinicalBERT models were evaluated, and model performance was assessed on 35 randomized subsets of the test set (each comprising 70% of test data). The Wilcoxon-Mann-Whitney test was used to evaluate statistical significance, with Bonferroni correction applied to control for multiple hypothesis testing. Baseline models included zero-shot and few-shot GPT-4o, Support Vector Machine, Random Forest, Logistic Regression, and Multinomial Naive Bayes.</jats:p>\n                  </jats:sec>\n                  <jats:sec>\n                    <jats:title>Results</jats:title>\n                    <jats:p>The proposed framework achieved the highest accuracy, with fully adapted Longformer reaching 89.83%. Performance improvements remained statistically significant after Bonferroni correction compared to all baselines (p &amp;lt; .05). The framework demonstrated robust gains across different stigmatizing language types.</jats:p>\n                  </jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion</jats:title>\n                    <jats:p>This study underscores the value of domain-adaptive NLP for detecting stigmatizing language in EHRs. The multi-stage transfer learning framework effectively captures subtle biases often missed by conventional models, enabling more objective and respectful clinical documentation.</jats:p>\n                  </jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion</jats:title>\n                    <jats:p>This framework offers a statistically validated, high-performing framework for detecting stigmatizing language in EHRs, supporting responsible AI and promoting equity in clinical care.</jats:p>\n                  </jats:sec>","is_dataset_classified":null,"base_score":1.3862943611198906,"endowment":1.3862943611198906,"datacite_reuse_total":0,"file_count":0,"downloads":0,"views":0,"has_version_chain":false,"is_dataset":false,"is_oa":false,"pmid":"41206907","pmcid":null,"openalex_id":"https://openalex.org/W4416053746","authors":[],"funders":[],"total_grants":0,"fwci":3.7644,"citation_percentile":0.93662648,"influential_citations":0,"citation_trend":[{"year":2025,"count":1},{"year":2026,"count":2}],"oa_status":"green","license":"https://academic.oup.com/pages/standard-publication-reuse-rights","oa_locations":[{"url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC12844570/","host_type":"repository"},{"url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC12844570/","host_type":"repository"},{"url":"https://academic.oup.com/jamia/article-pdf/33/2/283/65248079/ocaf193.pdf","host_type":"publisher"},{"url":"https://doi.org/10.1093/jamia/ocaf193","host_type":"journal"},{"url":"https://pubmed.ncbi.nlm.nih.gov/41206907","host_type":"repository"}],"fields_of_study":["Machine Learning in Healthcare","Topic Modeling","Artificial Intelligence in Healthcare and Education","Electronic Health Records","Humans","Machine Learning","Language","Natural Language Processing","Semantics","Social Stigma","Support Vector Machine"],"mesh_terms":["Machine Learning","Humans","Language","Natural Language Processing","Semantics","Electronic Health Records","Social Stigma","Support Vector Machine"],"keywords":["Health records","Transfer of learning","Electronic health record","Health equity","Equity (law)","Language understanding","Medical record","Electronic Health Records","Natural Language Processing","Stigmatizing Language","Transfer Learning","Large Language Models (Llms)"],"sdg_mappings":[],"linked_datasets":[],"clinical_trials":[],"software_tools":[],"database_accessions":[],"source":"live","citation_network_status":"fetched"},"created_at":"2026-08-04T01:22:27.826638Z","pmid":null,"pmcid":null,"fwci":null,"citation_percentile":null,"influential_citations":0,"oa_status":null,"license":null,"views":0,"total_file_size_bytes":0,"version_count":0,"fair_f":null,"fair_a":null,"fair_i":null,"fair_r":null,"fair_zscore":null,"fair_rationale":null,"fair_model":null,"fair_agent_version":null,"fair_fulltext_source":null,"fair_has_llm":null,"fair_computed_at":null,"clinical_trials":[],"software_tools":[],"db_accessions":[],"linked_datasets":[],"topics":[]}