{ "schema_version": "2.0", "name": "YLab Publications", "canonical_page": "https://ylab.top/publication/", "author_orcid": "https://orcid.org/0000-0001-5696-363X", "all_bibtex_url": "https://ylab.top/bib/publications.bib", "citation_sitemap_url": "https://ylab.top/citation-sitemap.xml", "number_of_publications": 77, "publications": [ { "title": "BRIDGE: Benchmarking Large Language Models for Understanding Real-World Clinical Practice Texts", "authors": [ "Jiageng Wu", "Bowen Gu", "Ren Zhou", "Kevin Xie", "Doug Snyder", "Yixing Jiang", "Valentina Carducci", "Richard Wyss", "Rishi J Desai", "Emily Alsentzer", "Leo Anthony Celi", "Adam Rodman", "Sebastian Schneeweiss", "Jonathan H Chen", "Santiago Romero-Brufau", "Kueiyu Joshua Lin", "Jie Yang" ], "year": 2026, "venue": "Nature Biomedical Engineering", "url": "https://www.nature.com/articles/s41551-026-01719-2", "citation_page_url": "https://ylab.top/citation/bridge-benchmarking-large-language-models-for-understanding-real-world-clinical-practice-texts/", "bibtex_url": "https://ylab.top/bib/bridge-benchmarking-large-language-models-for-understanding-real-world-clinical-practice-texts.bib", "citation_key": "Wu2026Bridgebenchmarking", "bibtex_type": "article", "doi": "10.1038/s41551-026-01719-2", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "code_url": "https://github.com/YLab-Open/BRIDGE", "data_url": "https://huggingface.co/datasets/YLab-Open/BRIDGE-Open", "preprint_url": "https://arxiv.org/abs/2504.19467", "publisher": "Springer Science and Business Media LLC", "abstract": "Large language models (LLMs) are evolving rapidly and hold great promise for medical applications, yet benchmarking on real-world clinical data such as electronic health records remains limited. Most existing benchmarks rely on medical examination-style questions or PubMed-derived text, failing to capture the complexity of clinical practice, while others target specific application scenarios with limited generalizability. Here we present BRIDGE, a comprehensive multilingual benchmark comprising 87 tasks sourced from 59 real-world clinical data sources across 9 languages. It covers eight task types spanning the patient care continuum, including triage, information extraction, diagnosis, prognosis and billing coding, and involves 14 clinical specialties. We systematically evaluated 95 LLMs (including DeepSeek-R1, GPT-4o, Gemini and Qwen3) under multiple inference strategies. Results reveal substantial performance variation across model sizes, languages, natural language processing tasks and clinical specialties. Open-source LLMs can match proprietary models, while medically fine-tuned models built on older backbones often lag behind updated general-purpose LLMs. BRIDGE and its continuously updated leaderboard provide foundational resources and important references for evaluating and developing LLMs for real-world clinical text understanding.", "published_url": "https://doi.org/10.1038/s41551-026-01719-2", "crossref_type": "journal-article", "pmid": "42310130", "published_version_url": "https://doi.org/10.1038/s41551-026-01719-2" }, { "title": "Large Language Models Generate Stigmatizing Language During Reasoning Over Real-World Clinical Data", "authors": [ "Yutong Yang", "Bowen Gu", "David B Hathaway", "Richard Wyss", "Laura Marengo", "Jason B Gibbons", "Stanley Lyndon", "Jiageng Wu", "Qingyu Chen", "Nan Liu", "Philip S Wang", "Leo Anthony Celi", "David W Bates", "Kueiyu Joshua Lin", "Li Zhou", "Jie Yang" ], "year": 2026, "venue": "medRxiv preprint", "url": "https://www.medrxiv.org/content/10.64898/2026.08.12.26360210v2", "citation_page_url": "https://ylab.top/citation/large-language-models-generate-stigmatizing-language-during-reasoning-over-real-world-clinical-data/", "bibtex_url": "https://ylab.top/bib/large-language-models-generate-stigmatizing-language-during-reasoning-over-real-world-clinical-data.bib", "citation_key": "Yang2026StigmatizingLanguage", "bibtex_type": "article", "doi": "10.64898/2026.08.12.26360210", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "code_url": "https://github.com/YLab-Open/LLM-Stigma-Language", "data_url": "https://github.com/YLab-Open/LLM-Stigma-Language", "preprint_url": "https://www.medrxiv.org/content/10.64898/2026.08.12.26360210v2", "publisher": "openRxiv", "abstract": "Stigmatizing language in clinical documentation, which conveys negative stereotypes, attitudes, or judgments toward patients, is a recognized source of documentation bias and is associated with poorer care and adverse health outcomes. Although prior stigma-related research has explored on clinician-written EHR notes, the increasing use of large language model (LLM)-generated documentation in clinical workflows raises new concerns about its potential to produce or amplify bias and affect patient safety. In this study, we conducted a large-scale assessment of stigmatizing language in LLM-generated reasoning text on 35 real-world clinical tasks across 107 LLMs. We applied a psychiatrist-validated, natural language processing (NLP) system to detect stigma terms in LLM reasoning text and quantified stigma rates of LLM-generated reasoning texts across 3,745 model-task pairs. Results showed that stigma rates ranged from 0% to 33.33%, with 84.06% of pairs containing stigma terms. Reasoning models showed higher stigma rates than non-reasoning models (2.35% vs. 1.70%; p < 0.0001), whereas medical models did not show significantly lower stigma rates than general-purpose models (1.80% vs. 2.00%; p = 0.26). Stigma rates of LLM outputs correlated negatively with task accuracy (r = -0.283; p < 0.001) and positively with input clinical-text stigma (r = 0.569; p < 0.001), with 19.76% of LLM-task pairs amplifying stigma in the original input notes. The effectiveness of prompt engineering as a destigmatizing approach varied across models, with stigma rates reduced by up to 91.91% without compromising model performance. This study shows that stigmatizing language generation is common but modifiable in LLM-generated reasoning traces, highlighting the need for direct stigma-related safety evaluation and robust destigmatization strategies before clinical deployment.", "published_url": "https://doi.org/10.64898/2026.08.12.26360210", "crossref_type": "posted-content", "preprint_version": "v2" }, { "title": "An Expedited Chart Review Process for Large Database Studies Using Natural Language Processing and Multiwave Adaptive Sampling", "authors": [ "Shirley V Wang", "Georg Hahn", "Sreedhara Sushama Kattinakere", "Mufaddal Mahesri", "Haritha S Pillai", "Rajendra Aldis", "Joyce Lii", "Sarah K Dutcher", "Rhoda Eniafe", "Jamal T Jones", "Keewan Kim", "Jiwei He", "Hana Lee", "Sengwee Toh", "Rishi J Desai", "Jie Yang" ], "year": 2026, "venue": "Epidemiology", "url": "https://pubmed.ncbi.nlm.nih.gov/41944789/", "citation_page_url": "https://ylab.top/citation/an-expedited-chart-review-process-for-large-database-studies-using-natural-language-processing-and-multiwave-adaptive-sampling/", "bibtex_url": "https://ylab.top/bib/an-expedited-chart-review-process-for-large-database-studies-using-natural-language-processing-and-multiwave-adaptive-sampling.bib", "citation_key": "Wang2026Expeditedchart", "bibtex_type": "article", "doi": "10.1097/ede.0000000000001978", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "37", "issue": "4", "pages": "504-514", "publisher": "Ovid Technologies (Wolters Kluwer Health)", "abstract": "Background One of the ways to enhance analyses conducted with large claims databases is by validating the measurement characteristics of the code-based algorithms used to identify health outcomes or other key study parameters of interest. These metrics can be used in quantitative bias analyses to assess the robustness of results for an inferential study, given potential bias from outcome misclassification. However, performing this validation through manual chart review of free-text notes from linked electronic health records requires extensive time and resource allocation. Methods We describe an expedited process for validating code-based algorithms that introduces efficiency using two distinct mechanisms: (1) use of natural language processing to reduce the time spent by human reviewers to review each chart, and (2) a multiwave adaptive sampling approach with predefined criteria to stop the validation study once performance characteristics are identified with sufficient precision. We illustrate this process in a case study that validates the performance of a claims-based outcome algorithm for intentional self-harm in patients with obesity. Results We empirically demonstrate that the natural language processing-assisted annotation process reduced the time spent on review per chart by 40%, and the use of the predefined stopping rule with multiwave samples would have prevented review of 77% of patient charts with limited compromise to the precision of performance characteristics. Conclusion This approach could facilitate more routine validation of code-based algorithms used to define key study parameters, ultimately enhancing understanding of the reliability of findings derived from database studies.", "published_url": "https://doi.org/10.1097/ede.0000000000001978", "crossref_type": "journal-article", "pmid": "41944789" }, { "title": "DT-Transformer: A Foundation Model for Disease Trajectory Prediction on a Real-World Health System", "authors": [ "Yunying Zhu", "Andrew R Weckstein", "Kueiyu Joshua Lin", "Jie Yang" ], "year": 2026, "venue": "arXiv preprint", "url": "https://arxiv.org/abs/2605.14227", "citation_page_url": "https://ylab.top/citation/dt-transformer-a-foundation-model-for-disease-trajectory-prediction-on-a-real-world-health-system/", "bibtex_url": "https://ylab.top/bib/dt-transformer-a-foundation-model-for-disease-trajectory-prediction-on-a-real-world-health-system.bib", "citation_key": "Zhu2026Dttransformer", "bibtex_type": "article", "doi": null, "metadata_sources": [ "arXiv" ], "preprint_url": "https://arxiv.org/abs/2605.14227", "arxiv_id": "2605.14227", "abstract": "Accurate disease trajectory prediction is critical for early intervention, resource allocation, and improving long-term outcomes. While electronic health records (EHRs) provide a rich longitudinal view of patient health in clinical environments, models trained on curated research cohorts may not reflect routine deployment settings, and those trained on single-hospital datasets capture only fragments of each patient's trajectory. This highlights the importance of leveraging large, multi-hospital health systems for training and validation to better reflect real-world clinical complexity. In this work, we develop DT-Transformer, a foundation model trained on 57.1M structured EHR entries over 1.7M patients from Mass General Brigham (MGB), spanning 11 hospitals and a broad network of outpatient clinics. DT-Transformer achieves strong discrimination in both held-out and prospective validation settings. Next-event prediction achieves a median age- and sex-stratified AUC of 0.871 across 896 disease categories, with all categories exceeding AUC 0.5. These results support health system-scale training as a path toward foundation models suited to real-world clinical forecasting." }, { "title": "Trustworthy Classification for Complex Social Surveys: A Memory-Enhanced Hierarchical Framework with Calibrated Uncertainty", "authors": [ "Zeqiang Wang", "Rebecca Oldroyd", "Yuqi Wang", "Jiageng Wu", "Jie Yang", "Wei Wang", "Nishanth R Sastry", "Jon Johnson", "Suparna De" ], "year": 2026, "venue": "Proceedings of the AAAI Conference on Artificial Intelligence", "url": "https://ojs.aaai.org/index.php/AAAI/article/view/41291", "citation_page_url": "https://ylab.top/citation/trustworthy-classification-for-complex-social-surveys-a-memory-enhanced-hierarchical-framework-with-calibrated-uncertainty/", "bibtex_url": "https://ylab.top/bib/trustworthy-classification-for-complex-social-surveys-a-memory-enhanced-hierarchical-framework-with-calibrated-uncertainty.bib", "citation_key": "Wang2026TrustworthyClassification", "bibtex_type": "inproceedings", "doi": "10.1609/aaai.v40i46.41291", "metadata_sources": [ "Crossref" ], "metadata_match": { "crossref": 1.0 }, "volume": "40", "issue": "46", "pages": "39415-39422", "publisher": "Association for the Advancement of Artificial Intelligence (AAAI)", "abstract": "Automated classification of complex social survey questionnaires is crucial for large-scale social science research but faces significant reliability challenges due to intricate hierarchical label structures, severe class imbalance, semantic ambiguity, and incomplete data coverage. Conventional classification methods often struggle with these combined complexities, yielding results that lack trustworthiness. We introduce HOCM, a framework designed for trustworthy classification in complex, real-world taxonomies. It features two synergistic components: (1) memory-enhanced contrastive learning, tailored to learn robust representations from noisy, imbalanced data by leveraging quality-aware category memory banks; and (2) hierarchical uncertainty calibration, which enforces taxonomic consistency while providing reliable confidence estimates and identifying inputs falling outside well-represented known categories. Our evaluation on a large-scale, real-world social survey dataset—a challenging exemplar of our target problem class—demonstrates that HOCM maintains strong accuracy on known classes while effectively identifying uncertain cases, significantly boosting accuracy on confident predictions. Furthermore, it adeptly detects low-resource/unknown categories. HOCM provides a more reliable automated classification tool, enabling efficient expert review and enhancing the trustworthiness of analysis in domains with complex, hierarchical data.", "published_url": "https://doi.org/10.1609/aaai.v40i46.41291", "crossref_type": "journal-article" }, { "title": "Linguistic Effects of Ambient AI on Clinical Documentation: A Matched Pre-Post Study", "authors": [ "Yiming Li", "Huiyuan Zhou", "Suzanne Blackley", "Joseph M Plasek", "Zhuoyang Lyu", "Wei Zhang", "Jacqueline You", "Amanda Centi", "Rebecca Mishuris", "Jie Yang", "Li Zhou" ], "year": 2026, "venue": "medRxiv preprint", "url": "https://www.medrxiv.org/content/10.64898/2026.02.16.26346370v1", "citation_page_url": "https://ylab.top/citation/linguistic-effects-of-ambient-ai-on-clinical-documentation-a-matched-pre-post-study/", "bibtex_url": "https://ylab.top/bib/linguistic-effects-of-ambient-ai-on-clinical-documentation-a-matched-pre-post-study.bib", "citation_key": "Li2026Linguisticeffects", "bibtex_type": "article", "doi": "10.64898/2026.02.16.26346370", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "preprint_url": "https://www.medrxiv.org/content/10.64898/2026.02.16.26346370v1", "publisher": "openRxiv", "abstract": "Ambient intelligence–based systems are increasingly used for clinical documentation. To quantify linguistic differences associated with ambient documentation, we conducted a matched pre–post analysis of 6,026 outpatient clinical notes from Mass General Brigham following implementation of two ambient AI documentation systems (Nuance Dragon Ambient eXperience [DAX] and Abridge). Within-clinician comparisons focused on the History of Present Illness (HPI) and Assessment and Plan (A&P) sections and evaluated syntactic complexity, lexical ambiguity, linguistic variability, discourse coherence, and readability. Manual review of 50 paired notes was performed to validate findings from automated linguistic analyses. Our analyses indicate that the linguistic effects of ambient documentation are both vendor-dependent and section-specific. Across both vendors, ambient notes in HPI were longer and exhibited greater syntactic complexity (longer sentences and clauses, increased dependency distance), lower lexical ambiguity, lower language-model perplexity, and higher local and global discourse coherence. These findings indicate that ambient systems systematically restructure conversational input into more syntactically elaborated and linguistically predictable narratives, reflecting increased standardization relative to both general-domain and biomedical language models. In contrast, changes in A&P were smaller and more heterogeneous, consistent with its more structured/templated nature. Readability analyses further showed increased length and lexical complexity in ambient HPI, whereas A&P readability differences were minimal. Overall, our findings demonstrate that ambient documentation changes how clinical information is linguistically expressed and organized, with effects varying by note section, vendor, and provider role/specialty. Evaluation should therefore extend beyond efficiency to consider effects on communication, cognitive load, clinical inference, and downstream analytics.", "published_url": "https://doi.org/10.64898/2026.02.16.26346370", "crossref_type": "posted-content", "preprint_version": "v1" }, { "title": "Transcranial Magnetic Stimulation Retreatment for Recurrent Depression: A Large Real-World Multisite Cohort Study", "authors": [ "Stanley Lyndon", "Jason B Gibbons", "Loreen Straub", "Meekang Sung", "Jacob C Jameson", "David B Hathaway", "Denise H Mulcahy", "Cason Elliott", "Rachel Wood", "Matthew Ruble", "Alejandro G Szmulewicz", "Jie Yang", "Rajendra Aldis", "Richard Wyss", "Philip S Wang", "Shan H Siddiqi" ], "year": 2026, "venue": "Acta Psychiatrica Scandinavica", "url": null, "citation_page_url": "https://ylab.top/citation/transcranial-magnetic-stimulation-retreatment-for-recurrent-depression-a-large-real-world-multisite-cohort-study/", "bibtex_url": "https://ylab.top/bib/transcranial-magnetic-stimulation-retreatment-for-recurrent-depression-a-large-real-world-multisite-cohort-study.bib", "citation_key": "Lyndon2026Transcranialmagnetic", "bibtex_type": "article", "doi": "10.1111/acps.70130", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 0.9955, "europe_pmc": 1.0 }, "pages": "acps.70130", "publisher": "Wiley", "abstract": "Introduction When depressive symptoms return after an initial response to transcranial magnetic stimulation (TMS), evidence to guide use of a second acute course is strikingly limited. We aimed to estimate how often and how quickly patients with recurrent major depressive disorder benefit from a second course of transcranial magnetic stimulation after they previously improved with TMS. Methods A retrospective cohort study across 20 outpatient centres included 259 patients who received ≥ 2 TMS courses between January 1, 2023 and October 1, 2025, and had clinically significant depressive symptoms at the start of course 2; 177 prior responders formed the primary cohort. Treatment used standard 10-Hz left dorsolateral prefrontal cortex TMS. Response and remission were assessed using the PHQ-9 and/or BDI-II; response was defined as a ≥ 50% reduction and remission as PHQ-9 ≤ 4 and/or BDI-II ≤ 9. Symptom change from baseline to end of course 2 and time to first response by session were compared between courses. Results Among past responders, 146 of 177 patients (82.5%; 95% CI, 76.1%-87.8%) responded again during course 2 and 96 of 177 (54.2%; 95% CI, 46.6%-61.7%) achieved remission. PHQ-9 scores decreased by a median of 65.0% (p Conclusions Most patients who previously benefited from acute TMS improved again when symptoms recurred, and among responders, the first session meeting response criteria often occurred earlier during retreatment than during the initial course.", "published_url": "https://doi.org/10.1111/acps.70130", "published_title": "Transcranial Magnetic Stimulation Retreatment for Recurrent Depression: A Large Real‐World Multisite Cohort Study", "crossref_type": "journal-article", "pmid": "42532518", "pmcid": "PMC13505613" }, { "title": "Predicting Deprescribing of High-Risk Medications Using Provider EHR Use and Patient Characteristics", "authors": [ "Bowen Gu", "Katharina Tabea Jungo", "Julie Lauffenburger", "Niteesh Choudhry", "Thomas Isaac", "John Zambrano", "Jie Yang" ], "year": 2026, "venue": "medRxiv preprint", "url": "https://www.medrxiv.org/content/10.64898/2026.07.25.26358746v1", "citation_page_url": "https://ylab.top/citation/predicting-deprescribing-of-high-risk-medications-using-provider-ehr-use-and-patient-characteristics/", "bibtex_url": "https://ylab.top/bib/predicting-deprescribing-of-high-risk-medications-using-provider-ehr-use-and-patient-characteristics.bib", "citation_key": "Gu2026Predictingdeprescribing", "bibtex_type": "article", "doi": "10.64898/2026.07.25.26358746", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "preprint_url": "https://www.medrxiv.org/content/10.64898/2026.07.25.26358746v1", "publisher": "openRxiv", "abstract": "Potentially inappropriate medications expose older adults to preventable harm, yet deprescribing remains difficult to implement consistently. Although electronic health record (EHR) interventions can reduce prescribing, health systems lack clear evidence about which routinely captured patient, primary care provider (PCP), and intervention-design factors predict medication discontinuation or dose tapering. Understanding these determinants is essential for targeting and scaling deprescribing support. In this study, we conducted the first machine-learning analysis of these trial data. We analyzed 2,979 adults aged 65 years or older and 158 structured EHR features spanning patient characteristics, PCP characteristics and EHR-use behaviors, and deprescribing-tool design. We compared eight models for predicting medication discontinuation or dose tapering and used SHAP to examine feature importance. TabPFN achieved the highest positive predictive value (67.71%), AUROC (74.30%), and AUPRC (60.95%), although overall predictability was moderate. Our findings show that even rich structured EHR data only moderately predict deprescribing, suggesting that important clinical determinants are not captured in routine fields. PCP EHR-use measures accounted for 19 of the 25 highest-ranked TabPFN features, although they also constituted most candidate predictors. The study provides the health system with an informative reference for predicting high-risk medication deprescribing. Future models should incorporate richer clinical context and undergo external validation before informing personalized deprescribing support.", "published_url": "https://doi.org/10.64898/2026.07.25.26358746", "crossref_type": "posted-content", "preprint_version": "v1" }, { "title": "Using Natural Language Processing of Clinical Notes to Supplement Structured Electronic Health Record Data for Phenotyping Smoking and Obesity in a Healthcare System", "authors": [ "Jie Yang", "Bowen Gu", "Haritha Pillai", "Joyce Lii", "David Cronkite", "Keith A Marsolo", "Rishi J Desai" ], "year": 2026, "venue": "Pharmacoepidemiology and Drug Safety", "url": "https://onlinelibrary.wiley.com/doi/full/10.1002/pds.70415", "citation_page_url": "https://ylab.top/citation/using-natural-language-processing-of-clinical-notes-to-supplement-structured-electronic-health-record-data-for-phenotyping-smoking-and-obesity-in-a-healthcare-system/", "bibtex_url": "https://ylab.top/bib/using-natural-language-processing-of-clinical-notes-to-supplement-structured-electronic-health-record-data-for-phenotyping-smoking-and-obesity-in-a-healthcare-system.bib", "citation_key": "Yang2026Naturallanguage", "bibtex_type": "article", "doi": "10.1002/pds.70415", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "preprint_url": "https://doi.org/10.64898/2026.01.18.26344356", "preprint_doi": "10.64898/2026.01.18.26344356", "volume": "35", "issue": "7", "pages": "e70415", "publisher": "Wiley", "abstract": "Purpose Studies based on electronic health records (EHR) often rely on structured data, which may incompletely capture important clinical phenotypes in EHR notes. The purpose of this study was to assess two natural language processing (NLP) tools to extract phenotypes from unstructured EHR notes, and to evaluate the added value of integrating NLP-derived phenotypes with structured EHR data at a health system scale. Methods This retrospective study is based on inpatient and outpatient EHR data from the Mass General Brigham healthcare system between January 1, 2019 and December 31, 2020. Two established rule-based NLP tools were applied to extract smoking and obesity information from 19 215 303 clinical notes of 503 025 patients. NLP performance was evaluated through a manual review of stratified samples. Phenotype prevalence was estimated using structured EHR data alone and compared with prevalence estimates obtained by supplementing structured data with NLP-derived features. Results Both NLP tools exhibited high performance, with both accuracy and the F1 score of 0.99 for smoking, and 0.92 and 0.91 for obesity, respectively. The combination of NLP and structured data identified 220 714 patients (43.88%) with smoking, compared with 170 396 patients (33.87%) identified using structured data alone, representing a 29.5% relative increase. For obesity, NLP identified 121 360 patients (24.13%) from EHR notes, and 169 905 patients (33.78%) were documented in structured data; the inclusion of NLP-derived features contributed an additional 32 823 patients, corresponding to a 19.3% relative increase. Conclusion NLP-derived phenotypes from unstructured EHR notes substantially improve patient identification for both smoking and obesity compared with structured EHR data alone at scale.", "published_url": "https://doi.org/10.1002/pds.70415", "crossref_type": "journal-article", "pmid": "42301002", "published_version_url": "https://doi.org/10.1002/pds.70415" }, { "title": "Scalable Medication Extraction and Discontinuation Identification from Electronic Health Records Using Large Language Models", "authors": [ "Chong Shao", "Douglas Snyder", "Chiran Li", "Bowen Gu", "Kerry Ngan", "Chun-Ting Yang", "Jiageng Wu", "Richard Wyss", "Kueiyu Joshua Lin", "Jie Yang" ], "year": 2026, "venue": "Journal of Clinical Epidemiology", "url": "https://pubmed.ncbi.nlm.nih.gov/41232578/", "citation_page_url": "https://ylab.top/citation/scalable-medication-extraction-and-discontinuation-identification-from-electronic-health-records-using-large-language-models/", "bibtex_url": "https://ylab.top/bib/scalable-medication-extraction-and-discontinuation-identification-from-electronic-health-records-using-large-language-models.bib", "citation_key": "Shao2026Scalablemedication", "bibtex_type": "article", "doi": "10.1016/j.jclinepi.2025.112049", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "pmid": "41232578", "volume": "189", "pages": "112049", "publisher": "Elsevier BV", "abstract": "Objectives Identifying medication discontinuations in electronic health records (EHRs) is vital for patient safety but is often hindered by information being buried in unstructured notes. This study aims to evaluate the capabilities of advanced open-sourced and proprietary large language models in extracting medications and classifying their medication status from EHR notes, focusing on their scalability for medication information extraction without human annotation. Study design and setting We collected three EHR datasets from diverse sources to build the evaluation benchmark: 1 publicly available dataset (Reannotated Clinical Acronym Sense Inventory dataset [Re-CASI]), 1 we annotated based on public MIMIC notes (MIMIC-IV Medication Snippet dataset [MIV-Med]), and 1 internally annotated on clinical notes from Mass General Brigham (MGB-Med). We evaluated 12 advanced LLMs, including general-domain open-sourced models (eg, Llama-3.1-70B-Instruct, Qwen2.5-72B-Instruct), medical-specific models (eg, MeLLaMA-70B-chat), and a proprietary model (GPT-4o). We explored multiple LLM prompting strategies, including zero-shot, 5-shot, and Chain-of-Thought (CoT) approaches. Performance on medication extraction, medication status classification, and their joint task (extraction then classification) was systematically compared across all experiments. Results LLMs showed promising performance on medication extraction, while discontinuation classification and joint tasks were more challenging. GPT-4o consistently achieved the highest average F1 scores in all tasks under zero-shot setting - 94.0% for medication extraction, 78.1% for discontinuation classification, and 72.7% for the joint task. Open-sourced models followed closely, with Llama-3.1-70B-Instruct achieving the highest performance in medication status classification on the MIV-Med dataset (68.7%) and in the joint task on both the Re-CASI (76.2%) and MIV-Med (60.2%) datasets. Medical-specific LLMs demonstrated lower performance compared to advanced general-domain LLMs. Few-shot learning generally improved performance, while CoT reasoning showed inconsistent gains. Notably, open-sourced models occasionally surpassed GPT-4o performance, underscoring their potential in privacy-sensitive clinical research. Conclusion LLMs demonstrate strong potential for medication extraction and discontinuation identification on EHR notes, with open-sourced models offering scalable alternatives to proprietary systems and few-shot learning further improving LLMs' capability. Plain language summary Stopping a medicine can affect safety and treatment decisions, yet this detail is often buried in long electronic health record notes. We evaluated whether large language models, which read and summarize text, can automatically find medication names and decide whether each medicine is still being taken, has been stopped, or neither. We tested 12 models, including open-source options suitable for secure hospital use, on three collections of clinical notes and compared three simple instruction styles: giving no examples, showing a few examples, and asking for step-by-step reasoning. All models produced usable results. The strongest systems scored about 94 for finding medication names and about 78 for deciding continued or stopped status, on a standard 0 to 100 measure that balances completeness and correctness. Showing a few examples usually helped more than step-by-step prompts, and several open-source models performed close to a leading proprietary system. These tools could help hospitals and researchers monitor medications at scale to support drug-safety studies, adherence tracking, and clinical decision support, with local validation and safeguards before clinical use.", "published_url": "https://doi.org/10.1016/j.jclinepi.2025.112049", "crossref_type": "journal-article", "pmcid": "PMC12714491" }, { "title": "Efficacy of Cognitive Behavioral Therapy-Based Smartphone App for Smoking Cessation in China: A Randomized Controlled Trial", "authors": [ "Jinsong Tang", "Xiaocong Liu", "Shanshan Chen", "Yunkai Sun", "Minzhe Zhu", "Zitang Zhou", "Yueheng Liu", "Jie Yang", "Wei Chen", "Yanhui Liao" ], "year": 2026, "venue": "Addiction", "url": "https://onlinelibrary.wiley.com/doi/10.1111/add.70239", "citation_page_url": "https://ylab.top/citation/efficacy-of-cognitive-behavioral-therapy-based-smartphone-app-for-smoking-cessation-in-china-a-randomized-controlled-trial/", "bibtex_url": "https://ylab.top/bib/efficacy-of-cognitive-behavioral-therapy-based-smartphone-app-for-smoking-cessation-in-china-a-randomized-controlled-trial.bib", "citation_key": "Tang2026Efficacycognitive", "bibtex_type": "article", "doi": "10.1111/add.70239", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 0.9959, "europe_pmc": 1.0 }, "volume": "121", "issue": "3", "pages": "575-585", "publisher": "Wiley", "abstract": "Background and aims Smoking is the leading preventable cause of premature death in China, yet less than 10% of smokers seeking to quit receive treatment. This study aimed to test the efficacy of a cognitive behavioral therapy (CBT)-based smoking cessation intervention ('Smokefree Life') via a smartphone app in China. Design A digital, parallel, single-blind, two-arm (1:1) randomized controlled trial. Setting This trial was conducted via 'Smokefree Life' app in the whole of mainland of China between 13 December 2021 and 15 April 2023. Participants Chinese-speaking adult smokers (925 males, 102 females, mean age 35.0 ± 6.7) willing to quit within one month were recruited via WeChat, survey sampling and referral from doctors, friends and family and randomized in a 1:1 ratio. Intervention The intervention group (n = 523) received a 13-14 week CBT-based smoking cessation intervention via smartphone app and the control group (n = 504) received only control messages. All participants were followed for 26 weeks post-quit date. Measurements The primary outcome was biochemically validated continuous abstinence at 26 weeks. Secondary outcomes were self-reported 7-day abstinence and continuous abstinence at 26 weeks. Findings By intention-to-treat analysis, the biochemically verified 26-week continuous abstinence rate was statistically significantly higher in the intervention group (6.50%) compared with the control group (2.58%) (odds ratio = 2.63, 95% confidence interval = 1.40-5.22, P = 0.0037). Self-reported 7-day and continuous abstinence rates were also consistently higher in the intervention group across follow-ups (all P Conclusions The 'Smokefree Life' app-based smoking cessation intervention statistically significantly improved 6-month biochemically confirmed quit rates among adult smokers in China.", "published_url": "https://doi.org/10.1111/add.70239", "published_title": "Efficacy of cognitive behavioral therapy‐based smartphone app for smoking cessation in China: A randomized controlled trial", "crossref_type": "journal-article", "pmid": "41218997", "pmcid": "PMC12887926" }, { "title": "MedKGent: A Large Language Model Agent Framework for Constructing Temporally Evolving Medical Knowledge Graph", "authors": [ "Duzhen Zhang", "Zixiao Wang", "Zhong-Zhi Li", "Yahan Yu", "Shuncheng Jia", "Jiahua Dong", "Haotian Xu", "Xing Wu", "Yingying Zhang", "Tielin Zhang", "Jie Yang", "Xiuying Chen", "Le Song" ], "year": 2026, "venue": "npj Digital Medicine", "url": "https://arxiv.org/abs/2508.12393", "citation_page_url": "https://ylab.top/citation/medkgent-a-large-language-model-agent-framework-for-constructing-temporally-evolving-medical-knowledge-graph/", "bibtex_url": "https://ylab.top/bib/medkgent-a-large-language-model-agent-framework-for-constructing-temporally-evolving-medical-knowledge-graph.bib", "citation_key": "Zhang2026Medkgentlarge", "bibtex_type": "article", "doi": "10.1038/s41746-026-03058-7", "metadata_sources": [ "Crossref", "arXiv" ], "metadata_match": { "crossref": 1.0 }, "preprint_url": "https://arxiv.org/abs/2508.12393", "publisher": "Springer Science and Business Media LLC", "abstract": "The rapid expansion of medical literature challenges the scalable structuring of domain knowledge. Knowledge Graphs (KGs) offer a solution, yet current construction methods lack generalizability and ignore the temporal dynamics of evolving knowledge. To address this, we introduce MedKGent, a Large Language Model (LLM) agent framework for building temporally evolving medical KGs. Using over 10 million PubMed abstracts from 1975 to 2023, MedKGent incrementally constructs a KG daily via two specialized agents. The Extractor Agent identifies knowledge triples and assigns confidence scores, while the Constructor Agent integrates these triples into a temporal graph, reinforcing recurring knowledge and resolving conflicts. The resulting KG contains 156,275 entities and 2,971,384 triples, making it, to our knowledge, the largest LLM-derived medical KG to date. Automated and expert assessments showed triple-validity rates approaching 90%. In downstream evaluations, MedKGent-KG significantly improved retrieval-augmented generation for five LLMs across seven medical question-answering benchmarks. Together, these results position MedKGent as a scalable and temporally aware infrastructure for medical knowledge representation and literature-grounded AI research.", "published_url": "https://doi.org/10.1038/s41746-026-03058-7", "crossref_type": "journal-article", "arxiv_id": "2508.12393", "published_version_url": "https://doi.org/10.1038/s41746-026-03058-7" }, { "title": "FinTruthQA: A Benchmark for AI-Driven Financial Disclosure Quality Assessment in Investor-Firm Interactions", "authors": [ "Peilin Zhou", "Ziyue Xu", "Xinyu Shi", "Jiageng Wu", "Yikang Jiang", "Dading Chong", "Wang Dong", "Jun Chen", "Bin Ke", "Jie Yang" ], "year": 2026, "venue": "Proceedings of the AAAI Symposium Series", "url": "https://ojs.aaai.org/index.php/AAAI-SS/article/view/42926", "citation_page_url": "https://ylab.top/citation/fintruthqa-a-benchmark-for-ai-driven-financial-disclosure-quality-assessment-in-investor-firm-interactions/", "bibtex_url": "https://ylab.top/bib/fintruthqa-a-benchmark-for-ai-driven-financial-disclosure-quality-assessment-in-investor-firm-interactions.bib", "citation_key": "Zhou2026FinTruthQA", "bibtex_type": "inproceedings", "doi": "10.1609/aaaiss.v9i1.42926", "metadata_sources": [ "Crossref" ], "metadata_match": { "crossref": 0.9953 }, "volume": "9", "issue": "1", "pages": "204-212", "publisher": "Association for the Advancement of Artificial Intelligence (AAAI)", "abstract": "Accurate and transparent financial information disclosure is essential for market efficiency, investor decision-making, and corporate governance. Chinese stock exchanges’ investor interactive platforms provide a widely used channel through which listed firms respond to investor concerns, yet these responses are often limited or non-substantive, making disclosure quality difficult to assess at scale. To address this challenge, we introduce FinTruthQA, to our knowledge the first benchmark for AI-driven assessment of financial disclosure quality in investor–firm interactions. FinTruthQA comprises 6,000 real-world financial Q&A entries, each manually annotated based on four key evaluation criteria: question identification, question relevance, answer readability, and answer relevance. We benchmark statistical machine learning models, pre-trained language models and their fine-tuned variants, as well as large language models (LLMs), on FinTruthQA. Experiments show that existing models achieve strong performance on question identification and question relevance (F1>95%), but remain substantially weaker on answer readability (Micro F1∼88%) and especially answer relevance (Micro F1∼80%), highlighting the nontrivial difficulty of fine-grained disclosure quality assessment. Domainand task-adapted pre-trained language models consistently outperform general-purpose models and LLM-based prompting on the most challenging settings. These findings position FinTruthQA as a practical foundation for AI-driven disclosure monitoring in capital markets, with value for regulatory oversight, investor protection, and disclosure governance in real-world financial settings.", "published_url": "https://doi.org/10.1609/aaaiss.v9i1.42926", "published_title": "FinTruthQA: A Benchmark for AI-Driven Financial Disclosure Quality Assessment in Investor–Firm Interactions", "crossref_type": "journal-article" }, { "title": "The Impact of Low EHR-Continuity on Effect Estimates: Evidence from Two EHR-Medicare Linked Databases", "authors": [ "Yinzhu Jin", "Angela Y Tong", "Richard Wyss", "Jie Yang", "Kueiyu Joshua Lin" ], "year": 2025, "venue": "Clinical Epidemiology", "url": null, "citation_page_url": "https://ylab.top/citation/the-impact-of-low-ehr-continuity-on-effect-estimates-evidence-from-two-ehr-medicare-linked-databases/", "bibtex_url": "https://ylab.top/bib/the-impact-of-low-ehr-continuity-on-effect-estimates-evidence-from-two-ehr-medicare-linked-databases.bib", "citation_key": "Jin2025Impactlow", "bibtex_type": "article", "doi": "10.2147/clep.s549772", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "pmid": "41473209", "pmcid": "PMC12746745", "volume": "17", "pages": "1131-1145", "publisher": "Informa UK Limited", "abstract": "Purpose To compare the effect estimates obtained from patients with different levels of electronic health record (EHR)-continuity in four empirical studies: risk of pneumonia among 1) new users of proton pump inhibitors (PPI) vs H2 receptor antagonists, or 2) new users of PPI vs non-PPI users; risk of major bleeding among 3) new users of warfarin vs direct-acting oral anticoagulants, or 4) new users of oral anti-coagulants (OAC) vs non-OAC users. Patients and methods Patients were identified in 2 US EHR systems (MA system, NC system) linked with Medicare claims data (2007/1/1 - 2014/12/31) separately. We calculated incidence rates (IR), incidence rate differences (IRD), and hazard ratios (HR) in the total linked study population and after excluding patients with the lowest 25%, 50%, or 75% of EHR-continuity scores. We quantified bias in IRD and propensity score (PS) decile-adjusted HR. Results In the MA system, IRs based on EHR-only data underestimated true rates by 44.1% to 76.2%, reduced to 12.9% to 46.5%. After excluding the lowest 75% of EHR-continuity patients, underestimation was more pronounced in non-user comparator designs. Absolute IRD bias was small for PPI vs H2RA (0.4%) and warfarin vs DOAC (0.7%), but larger for PPI vs non-PPI (19.1%) and OAC vs non-OAC (7.8%). Relative HR bias was 13.3% (PPI vs H2RA), 18.9% (PPI vs non-PPI), 3.0% (warfarin vs DOAC), and 31.5% (OAC vs non-OAC). Excluding lower-continuity patients and PS adjustment reduced IRD bias, while restricting to higher-continuity patients modestly improved HR bias. Conclusion Limiting analyses to patients with higher EHR-continuity can reduce IR underestimation and bias in effect estimates, particularly on the IRD scale. While PS adjustment mitigates some bias, EHR-discontinuity remains a source of bias, especially in studies using non-user comparators. These findings underscore the importance of balancing EHR-continuity with sample size considerations in pharmacoepidemiologic research.", "published_url": "https://doi.org/10.2147/clep.s549772", "crossref_type": "journal-article" }, { "title": "Validation of Treatment Discontinuation in Claims Data Using NLP and Electronic Health Records", "authors": [ "Chun-Ting Yang", "Kerry Ngan", "Dae Hyun Kim", "Jie Yang", "Jun Liu", "Kueiyu Joshua Lin" ], "year": 2025, "venue": "Innovation in Aging", "url": null, "citation_page_url": "https://ylab.top/citation/validation-of-treatment-discontinuation-in-claims-data-using-nlp-and-electronic-health-records/", "bibtex_url": "https://ylab.top/bib/validation-of-treatment-discontinuation-in-claims-data-using-nlp-and-electronic-health-records.bib", "citation_key": "Yang2025Validationtreatment", "bibtex_type": "article", "doi": "10.1093/geroni/igaf122.356", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "9", "issue": "Supplement_2", "pages": "igaf122.356", "publisher": "Oxford University Press (OUP)", "abstract": "Measuring medication discontinuation in claims data primarily relies on the gaps between prescription fills, but such definitions are rarely validated. This study aimed to establish a natural language processing (NLP)-based framework to validate claims-based discontinuation algorithms for commonly used medications against NLP-based reference standards from electronic health records (EHRs). We identified 35,010 patients receiving antipsychotic medications (APMs), benzodiazepines, warfarin, or direct oral anticoagulants (DOACs) from EHRs at the Mass General Brigham in 2007-2020. These EHR data were linked with 82,455 Medicare Part D claims. An NLP-aided chart review was applied to determine medication discontinuation from EHRs (reference standard). In claims data, we defined discontinuation based on a prescription gap larger than 15-90 days (claims-based algorithms). Sensitivity, specificity, and predictive values of claims-based algorithms against the reference standard were calculated. We found that the sensitivity and specificity of 90-day-gap-based algorithms were 0.49 and 0.82 for haloperidol, 0.47 and 0.86 for atypical APMs, 0.54 and 0.75 for benzodiazepines, 0.40 and 0.81 for warfarin, and 0.46 and 0.89 for DOACs, respectively. The corresponding estimates for 15-day-gap-based algorithms were 0.71 and 0.58 for haloperidol, 0.61 and 0.64 for atypical APMs, 0.75 and 0.45 for benzodiazepines, 0.64 and 0.50 for warfarin, and 0.61 and 0.66 for DOACs, respectively. Positive predictive values were primarily affected by medication discontinuation rates and less by gap lengths. In summary, the accuracy of claims-based discontinuation algorithms differs by medications and prescription gaps. This study demonstrates the scalability and utility of the NLP-based validation framework for multiple medications.", "published_url": "https://doi.org/10.1093/geroni/igaf122.356", "crossref_type": "journal-article", "pmcid": "PMC12759550" }, { "title": "Clinical Pathway-Aware Large Language Models for Reliable and Transparent Medical Dialogue", "authors": [ "Jiageng Wu", "Xian Wu", "Yefeng Zheng", "Jie Yang" ], "year": 2025, "venue": "Journal of Biomedical Informatics", "url": "https://arxiv.org/abs/2403.06611", "citation_page_url": "https://ylab.top/citation/clinical-pathway-aware-large-language-models-for-reliable-and-transparent-medical-dialogue/", "bibtex_url": "https://ylab.top/bib/clinical-pathway-aware-large-language-models-for-reliable-and-transparent-medical-dialogue.bib", "citation_key": "Wu2025Clinicalpathway", "bibtex_type": "article", "doi": "10.1016/j.jbi.2025.104942", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "preprint_url": "https://arxiv.org/abs/2403.06611", "volume": "172", "pages": "104942", "publisher": "Elsevier BV", "abstract": "Objective Large language models (LLMs) offer promising potential in answering real-time medical queries, but they often produce lengthy, generic, and even hallucinatory responses. We aim to develop a reliable and interpretable medical dialogue system that incorporates clinical reasoning and then mitigates the risk of hallucination. Methods Two large datasets of real-world online consultation, MedDG and KaMed, were used for evaluation. We proposed a Medical Dialogue System with Knowledge Enhancement and Clinical Pathway Encoding (MedKP), which integrates an external medical knowledge graph and encodes internal clinical pathways to model physician reasoning. Performance was compared with state-of-the-art baselines, including GPT-4o and LLaMA3.1-70B. A multi-dimensional evaluation framework assessed (1) clinical relevance (medical entity-based), (2) textual similarity (ROUGE, BLEU), (3) semantic alignment (BERTScore), and (4) hallucination and consistency via an external LLM-based judge, as well as parallel human evaluation. Results Across both datasets, MedKP (6B) achieved the best overall performance, outperforming other advanced baselines and producing responses that align more closely with those of human physicians. For clinical relevance, MedKP reached a macro F1-score of medical entity at 31.41 on MedDG (previous best DFMed: 24.76, improved 30.41%) and 26.62 on KaMed (previous best LLaM-A3.1-70B: 20.67, improved 25.62%). Consistent improvements were observed across other metrics. Ablation studies further validated the effectiveness of each model component. Conclusion Our results highlight the critical role of clinical reasoning in advancing trustworthy AI for digital healthcare. By enhancing the reliability, coherence, and transparency of AI-generated responses, this pathway-aware approach bridges the gap between LLMs and real-world clinical workflows, improving the accessibility of high-quality telemedicine services, particularly benefiting underserved populations.", "published_url": "https://doi.org/10.1016/j.jbi.2025.104942", "crossref_type": "journal-article", "pmid": "41177245", "arxiv_id": "2403.06611", "published_version_url": "https://doi.org/10.1016/j.jbi.2025.104942" }, { "title": "Why Chain of Thought Fails in Clinical Text Understanding", "authors": [ "Jiageng Wu", "Kevin Xie", "Bowen Gu", "Nils Krüger", "Kueiyu Joshua Lin", "Jie Yang" ], "year": 2025, "venue": "arXiv preprint", "url": "https://arxiv.org/abs/2509.21933", "citation_page_url": "https://ylab.top/citation/why-chain-of-thought-fails-in-clinical-text-understanding/", "bibtex_url": "https://ylab.top/bib/why-chain-of-thought-fails-in-clinical-text-understanding.bib", "citation_key": "Wu2025Whychain", "bibtex_type": "article", "doi": null, "metadata_sources": [ "arXiv" ], "preprint_url": "https://arxiv.org/abs/2509.21933", "arxiv_id": "2509.21933", "abstract": "Large language models (LLMs) are increasingly being applied to clinical care, a domain where both accuracy and transparent reasoning are critical for safe and trustworthy deployment. Chain-of-thought (CoT) prompting, which elicits step-by-step reasoning, has demonstrated improvements in performance and interpretability across a wide range of tasks. However, its effectiveness in clinical contexts remains largely unexplored, particularly in the context of electronic health records (EHRs), the primary source of clinical documentation, which are often lengthy, fragmented, and noisy. In this work, we present the first large-scale systematic study of CoT for clinical text understanding. We assess 95 advanced LLMs on 87 real-world clinical text tasks, covering 9 languages and 8 task types. Contrary to prior findings in other domains, we observe that 86.3\\% of models suffer consistent performance degradation in the CoT setting. More capable models remain relatively robust, while weaker ones suffer substantial declines. To better characterize these effects, we perform fine-grained analyses of reasoning length, medical concept alignment, and error profiles, leveraging both LLM-as-a-judge evaluation and clinical expert evaluation. Our results uncover systematic patterns in when and why CoT fails in clinical contexts, which highlight a critical paradox: CoT enhances interpretability but may undermine reliability in clinical text tasks. This work provides an empirical basis for clinical reasoning strategies of LLMs, highlighting the need for transparent and trustworthy approaches." }, { "title": "Generative Artificial Intelligence for Fundus Fluorescein Angiography Interpretation and Human Expert Evaluation", "authors": [ "An Shao", "Xiaocong Liu", "Wenyue Shen", "Yingyu Li", "Hongkang Wu", "Xiangji Pan", "Zexin Yang", "Yufeng Xu", "Tiepei Zhu", "Yao Wang", "Jie Yang", "Yih Chung Tham", "Jian Wu", "Kai Jin", "Juan Ye" ], "year": 2025, "venue": "npj Digital Medicine", "url": "https://www.nature.com/articles/s41746-025-01234-5", "citation_page_url": "https://ylab.top/citation/generative-artificial-intelligence-for-fundus-fluorescein-angiography-interpretation-and-human-expert-evaluation/", "bibtex_url": "https://ylab.top/bib/generative-artificial-intelligence-for-fundus-fluorescein-angiography-interpretation-and-human-expert-evaluation.bib", "citation_key": "Shao2025Generativeartificial", "bibtex_type": "article", "doi": "10.1038/s41746-025-01759-z", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "8", "issue": "1", "pages": "396", "publisher": "Springer Science and Business Media LLC", "abstract": "Fundus fluorescein angiography (FFA) is the gold standard for diagnosing chorioretinal diseases, but its interpretation requires significant expertise and time. Despite generative AI's enormous potential in medical report generation, automatic FFA interpretation lacks robust models and sufficient evaluation metrics. This study introduces InterpreFFA, a diagnosis-supervised contrastive learning framework, to emulate ophthalmologists' decision-making process in FFA report generation. Validated on multi-center datasets, InterpreFFA demonstrated superior performance and generalization compared to baseline models. In a simulated clinical setting, two residents used InterpreFFA to diagnose and report FFA cases, with six board-certified ophthalmologists rating the generated reports based on a five-point Likert scale. InterpreFFA significantly improved diagnostic accuracy (85.55 to 90.34%, p < 0.05) and shortened reporting time (153.93 to 108.08 s, p < 0.001). Although AI-generated reports scored slightly lower than manual reports (4.12 vs. 4.38, p < 0.01), InterpreFFA proves to be a promising and cost-effective ancillary tool for enhancing clinical efficiency.", "published_url": "https://doi.org/10.1038/s41746-025-01759-z", "crossref_type": "journal-article", "pmid": "40603524", "pmcid": "PMC12222453" }, { "title": "Comparative Ranking of Marginal Confounding Impact of Natural Language Processing-Derived versus Structured Features in Pharmacoepidemiology", "authors": [ "Joseph M Plasek", "Richard D Wyss", "Janick G Weberpals", "Jie Yang", "Thomas Deramus", "Theodore N Tsacogianis", "Kerry Ngan", "Lily G Bessette", "Kueiyu Joshua Lin", "Li Zhou" ], "year": 2025, "venue": "Computers in Biology and Medicine", "url": null, "citation_page_url": "https://ylab.top/citation/comparative-ranking-of-marginal-confounding-impact-of-natural-language-processing-derived-versus-structured-features-in-pharmacoepidemiology/", "bibtex_url": "https://ylab.top/bib/comparative-ranking-of-marginal-confounding-impact-of-natural-language-processing-derived-versus-structured-features-in-pharmacoepidemiology.bib", "citation_key": "Plasek2025Comparativeranking", "bibtex_type": "article", "doi": "10.1016/j.compbiomed.2025.109855", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "188", "pages": "109855", "publisher": "Elsevier BV", "abstract": "Objective To explore the ability of natural language processing (NLP) methods to identify confounder information beyond what can be identified using claims codes alone for pharmacoepidemiology. Methods We developed a retrospective cohort for high vs low dose proton pump inhibitors from linked Medicare claims (2008-2017) and electronic health record data for patients with a history of peptic ulcer disease or osteoarthritis. Clinical notes authored one year before first dispensing date were processed by off-the-shelf tools: bag-of-n-grams, latent Dirichlet allocation, a linguistics-focused tool, BERT sentence embeddings, BioBERT word embeddings, and GloVe word embeddings. Candidate features were ranked using Bross formula, a simple way to rank the marginal confounding impact of binary features on estimated causal effects. Results The marginal confounding impact in the Bross rankings of NLP-derived features trended from 39 % in the top 100 to 77 % in the top 500 to 93 % in the top 5000 among patients with peptic ulcer disease. More specifically, the top 25 confounders are largely from factors identified by domain experts and structured fields, and the marginal impact of these confounders is stronger than others. Features 25 to 50 include features identified by a linguistics-focused tool and embeddings, whereas features 50 to 100 include more embeddings and bag-of-ngrams. After 100, the curve flattens, meaning that the marginal impact of those potential confounders gets smaller. Similarly, among patients with osteoarthritis, NLP-derived features trended from 66 % in the top 100 to 84 % in the top 500 to 95 % in the top 5000 when the outcome was gastrointestinal bleed and from 47 % in the top 100 to 81 % in the top 500 to 94 % in the top 5000 when the outcome was acute kidney injury. Similar trends were observed in the information gain data, though NLP-derived features had higher baselines. Conclusions NLP contributed to finding large numbers of features that can supplement claims data and prespecified variables to help provide additional confounder information. We found that unsupervised off-the-shelf NLP tools can scale to generate large numbers of features appropriate for high-dimensional proxy adjustment and pharmacoepidemiology use cases.", "published_url": "https://doi.org/10.1016/j.compbiomed.2025.109855", "crossref_type": "journal-article", "pmid": "39965395" }, { "title": "Electronic Health Record (EHR) Enhanced Signal Detection Using Tree-Based Scan Statistic Methods", "authors": [ "Massimiliano Russo", "Sushama Kattinakere Sreedhara", "Joshua Smith", "Sharon E Davis", "Judith C Maro", "Thomas Deramus", "Joyce Lii", "Jie Yang", "Rishi Desai", "José J Hernández-Muñoz" ], "year": 2025, "venue": "American Journal of Epidemiology", "url": null, "citation_page_url": "https://ylab.top/citation/electronic-health-record-ehr-enhanced-signal-detection-using-tree-based-scan-statistic-methods/", "bibtex_url": "https://ylab.top/bib/electronic-health-record-ehr-enhanced-signal-detection-using-tree-based-scan-statistic-methods.bib", "citation_key": "Russo2025Electronichealth", "bibtex_type": "article", "doi": "10.1093/aje/kwaf199", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 0.9783, "crossref": 0.9783 }, "pmid": "40916726", "volume": "195", "issue": "1", "pages": "178-187", "publisher": "Oxford University Press (OUP)", "abstract": "Tree-based scan statistics (TBSS) are data mining methods that screen thousands of hierarchically related health outcomes to detect unsuspected adverse drug effects. TBSS traditionally analyze claims data with outcomes defined via diagnosis codes. TBSS have not been previously applied to rich clinical information in electronic health records (EHR). We developed approaches for integrating EHR data in TBSS analyses, including outcomes derived from natural language processing (NLP) applied to clinical notes and laboratory results, related via multipath hierarchical structures. We consider 4 settings that sequentially add sources of outcomes to the TBSS tree: (1) diagnosis code, (2) NLP-derived outcomes, (3) binary outcomes from lab results, and (4) continuous lab results. In a comparative cohort study involving second-generation sulfonylureas (SUs) and dipeptidyl peptidase 4 (DPP-4) inhibitors among adults with type 2 diabetes, with an a priori expected signal of hypoglycemia, diagnosis code data showed no statistical alerts for inpatient or emergency department settings. Adding NLP-derived outcomes resulted in an alert for “Headaches” (P = .047), a nonspecific symptom of hypoglycemia. Progressively adding binary and continuous lab results produced the same alert. Integrating EHR in TBSS can be useful for the detection of safety signals for further investigation.", "published_url": "https://doi.org/10.1093/aje/kwaf199", "published_title": "Electronic health record-enhanced signal detection using tree-based scan statistic methods", "published_year": 2026, "crossref_type": "journal-article" }, { "title": "Cognitive Assessment of Dementia Patients Using Large Language Models", "authors": [ "Jiageng Wu", "Richard Wyss", "Kueiyu Joshua Lin", "Jie Yang" ], "year": 2025, "venue": "Alzheimer's & Dementia", "url": null, "citation_page_url": "https://ylab.top/citation/cognitive-assessment-of-dementia-patients-using-large-language-models/", "bibtex_url": "https://ylab.top/bib/cognitive-assessment-of-dementia-patients-using-large-language-models.bib", "citation_key": "Wu2025Cognitiveassessment", "bibtex_type": "article", "doi": "10.1002/alz70858_103769", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "21", "issue": "S4", "pages": "e103769", "publisher": "Wiley", "abstract": "Background Cognitive function is a key determinant of daily functioning and quality of life in dementia patients. Accurate and timely cognitive assessments are critical for disease management, risk stratification, and effective intervention implementation. However, traditional assessments often require significant resources, limiting their scalability. Large language models (LLMs) offer a novel approach to predicting cognitive function by leveraging unstructured data from Electronic Health Records (EHRs). Method This retrospective cohort study analyzed EHRs of patients aged ≥65 years diagnosed with dementia between 2013 and 2020 at Mass General Brigham (MGB). Cognitive function was evaluated using the Brief Interview for Mental Status (BIMS) from the Minimum Data Set (MDS) or M1700 Cognitive Function from the Outcome and Assessment Information Set (OASIS). The clinical note from the past year were used as model input. Multiple state‐of‐the‐art LLMs were tested, including QWEN2.5 (7B and 72B), Llama 3.1 (8B), Llama 3.3 (70B), Ministral‐8B, Mistral Large, and DeepSeek‐Llama (70B). Three inference strategies were explored: (1) Standard mode, directly outputting predictions; (2) Chain‐of‐Thought (CoT) mode, generating step‐by‐step reasoning; and (3) Summary‐based mode, summarizing clinical notes before integrated prediction. Model performance was evaluated using F1‐scores with bootstrapping. Result The study included 685 patients who underwent MDS‐cognitive and 3,630 patients assessed using OASIS. Overall, LLMs demonstrated strong cross‐dataset generalizability. Among them, Llama‐3.3‐70B achieved the highest F1‐score (66.31%) in OASIS assessments, while QWEN2.5‐72B performed best in MDS (45.23%). The larger LLMs provided significant performance gains, whereas the CoT and summary strategies enhanced interpretability but did not consistently improve accuracy. This may be due to the models' limited understanding of clinical text, making precise information extraction and reasoning challenging. Conclusion This study highlights the potential of LLMs in predicting cognitive function for dementia patients using EHRs. Larger LLMs showed superior performance, while the CoT and summary strategies improved interpretability but may slightly impair performance. The results highlight the necessity for further adoption and optimization of LLM modeling and reasoning strategies to better handle clinical text, minimizing hallucinations and improving reliability. Integrating LLMs into cognitive assessment workflows could enhance efficiency and accessibility, supporting more effective dementia care and management.", "published_url": "https://doi.org/10.1002/alz70858_103769", "crossref_type": "journal-article", "pmcid": "PMC12740986" }, { "title": "Large Language Model for Predicting Altered Mental Status in Dementia Patients Using Electronic Health Records", "authors": [ "Jiageng Wu", "Richard Wyss", "Kueiyu Joshua Lin", "Jie Yang" ], "year": 2025, "venue": "Alzheimer's & Dementia", "url": null, "citation_page_url": "https://ylab.top/citation/large-language-model-for-predicting-altered-mental-status-in-dementia-patients-using-electronic-health-records/", "bibtex_url": "https://ylab.top/bib/large-language-model-for-predicting-altered-mental-status-in-dementia-patients-using-electronic-health-records.bib", "citation_key": "Wu2025Largelanguage", "bibtex_type": "article", "doi": "10.1002/alz70858_104165", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "21", "issue": "S4", "pages": "e104165", "publisher": "Wiley", "abstract": "Background Older adults with dementia face substantial challenges to their mental status due to declining physical function and deteriorating lifestyle, which negatively impacts their quality of life. Mental status is a key determinant of daily functioning in progressive dementia, but assessments often require substantial resources and may be inaccessible to many. Leveraging Electronic Health Records (EHRs) and large language models (LLMs) to predict mental status offers a promising alternative for improving patient management, risk assessment, and timely interventions. Method This retrospective cohort study analyzed EHRs of patients aged ≥65 years who were diagnosed with dementia between 2013 and 2020 and underwent Minimum Data Set (MDS) assessments at Mass General Brigham (MGB). Historical HER notes within the past year were utilized as model input. Altered mental status was defined based on the presence of C1310A Acute Mental Status Change or C1310 Signs and Symptoms of Delirium, including inattention, disorganized thinking, or altered level of consciousness. Three open‐source LLMs—QWEN2.5‐7B‐Instruct, Ministral‐8B‐Instruct, and Llama‐3.1‐8B‐Instruc were evaluated using two prompting strategies: a standard QA mode (direct answer output) and a Chain‐of‐Thought (CoT) mode (step‐by‐step reasoning). Performance was measured by accuracy using 1,000 bootstrapped iterations. Result The study included 779 patients (median age: 84 years, 44.2% male) and their corresponding 56,428 clinical notes from the past year, extracted from EHRs. Among the models, Llama‐3.1‐8B‐Instruct achieved the highest accuracy of 74.58% (95% CI: [74.48, 74.68]) in the standard QA mode. The CoT strategy improved reasoning transparency and interpretability but led to a slight performance decline compared to the standard QA approach. Conclusion This study demonstrates the potential of using EHRs and LLMs to predict altered mental status in dementia patients. The LLMs showed fine predictive performance, while the CoT strategy provided enhanced interpretability at the cost of reduced accuracy. These findings emphasize the need to optimize LLM in clinical applications and highlight the promise of advanced AI tools in improving dementia care through efficient risk assessment and timely interventions.", "published_url": "https://doi.org/10.1002/alz70858_104165", "crossref_type": "journal-article", "pmcid": "PMC12742264" }, { "title": "Establishing a Validation Framework of Treatment Discontinuation in Claims Data Using Natural Language Processing and Electronic Health Records", "authors": [ "Chun-Ting Yang", "Kerry Ngan", "Dae Hyun Kim", "Jie Yang", "Jun Liu", "Kueiyu Joshua Lin" ], "year": 2025, "venue": "Clinical Pharmacology & Therapeutics", "url": null, "citation_page_url": "https://ylab.top/citation/establishing-a-validation-framework-of-treatment-discontinuation-in-claims-data-using-natural-language-processing-and-electronic-health-records/", "bibtex_url": "https://ylab.top/bib/establishing-a-validation-framework-of-treatment-discontinuation-in-claims-data-using-natural-language-processing-and-electronic-health-records.bib", "citation_key": "Yang2025Establishingvalidation", "bibtex_type": "article", "doi": "10.1002/cpt.3650", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "118", "issue": "1", "pages": "138-145", "publisher": "Wiley", "abstract": "Measuring medication discontinuation in claims data primarily relies on the gaps between prescription fills, but such definitions are rarely validated. This study aimed to establish a natural language processing (NLP)-based validation framework to evaluate the performance of claims-based discontinuation algorithms for commonly used medications against NLP-based reference standards from electronic health records (EHRs). A total of 36,656 patients receiving antipsychotic medications (APMs), benzodiazepines (BZDs), warfarin, or direct oral anticoagulants (DOACs) were identified from the Mass General Brigham EHRs in 2007-2020. These EHR data were linked with 97,900 Medicare Part D claims. An NLP-aided chart review was applied to determine medication discontinuation from EHR (reference standard). In claims data, discontinuation was defined by having a prescription gap larger than 15-90 days (claims-based algorithms). Sensitivity, specificity, and predictive values of claims-based algorithms against the reference standard were measured. The sensitivity and specificity of 90-day-gap-based algorithms were 0.46 and 0.79 for haloperidol, 0.41 and 0.85 for atypical APMs, 0.47 and 0.75 for BZDs, 0.33 and 0.80 for warfarin, and 0.38 and 0.87 for DOACs, respectively. The corresponding estimates for 15-day-gap-based algorithms were 0.68 and 0.55 for haloperidol, 0.59 and 0.62 for atypical APMs, 0.71 and 0.45 for BZDs, 0.61 and 0.49 for warfarin, and 0.58 and 0.64 for DOACs, respectively. Positive predictive values were primarily affected by medication discontinuation rates and less by gap lengths. The overall accuracy of claims-based discontinuation algorithms differs by medications. This study demonstrates the scalability and utility of the NLP-based validation framework for multiple medications.", "published_url": "https://doi.org/10.1002/cpt.3650", "crossref_type": "journal-article", "pmid": "40197528", "pmcid": "PMC12167144" }, { "title": "Natural Language Processing for Scalable Feature Engineering and Ultra-High-Dimensional Confounding Adjustment in Healthcare Database Studies", "authors": [ "Richard Wyss", "Jie Yang", "Sebastian Schneeweiss", "Joseph M Plasek", "Li Zhou", "Thomas Deramus", "Janick G Weberpals", "Kerry Ngan", "Kueiyu Joshua Lin" ], "year": 2025, "venue": "Journal of Biomedical Informatics", "url": null, "citation_page_url": "https://ylab.top/citation/natural-language-processing-for-scalable-feature-engineering-and-ultra-high-dimensional-confounding-adjustment-in-healthcare-database-studies/", "bibtex_url": "https://ylab.top/bib/natural-language-processing-for-scalable-feature-engineering-and-ultra-high-dimensional-confounding-adjustment-in-healthcare-database-studies.bib", "citation_key": "Wyss2025Naturallanguage", "bibtex_type": "article", "doi": "10.1016/j.jbi.2025.104882", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "169", "pages": "104882", "publisher": "Elsevier BV", "abstract": "Background To improve confounding control in healthcare database studies, data-driven algorithms may empirically identify and adjust for large numbers of pre-exposure variables that indirectly capture information on unmeasured confounding factors ('proxy' confounders). Current approaches for high-dimensional proxy adjustment do not leverage free-text notes from electronic health records (EHRs). Unsupervised natural language processing (NLP) technology can scale to generate large numbers of structured features from unstructured notes. Objective To assess the impact of supplementing claims data analyses with large numbers of NLP generated features for high-dimensional proxy adjustment. Methods We linked Medicare claims with EHR data to generate three cohorts comparing different classes of medications on the 6-month risk of cardiovascular outcomes. We used various NLP methods to generate structured features from free-text EHR notes and used least absolute shrinkage and selection operator (LASSO) regression to fit several propensity score (PS) models that included different covariate sets as candidate predictors. Covariate sets included features generated from claims data only, and claims data plus NLP-generated EHR features. Results Including both claims codes and NLP-generated EHR features as candidate predictors improved overall covariate balance with standardized differences being Conclusion Supplementing administrative claims with large numbers of NLP-generated features for ultra-high-dimensional proxy confounder adjustment improved overall covariate balance and may provide a modest benefit in terms of capturing confounder information.", "published_url": "https://doi.org/10.1016/j.jbi.2025.104882", "crossref_type": "journal-article", "pmid": "40691893" }, { "title": "Scalable Information Extraction from Free Text Electronic Health Records Using Large Language Models", "authors": [ "Bowen Gu", "Vivian Shao", "Ziqian Liao", "Valentina Carducci", "Santiago Romero-Brufau", "Jie Yang", "Rishi Desai" ], "year": 2025, "venue": "BMC Medical Research Methodology", "url": "https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-024-02411-2", "citation_page_url": "https://ylab.top/citation/scalable-information-extraction-from-free-text-electronic-health-records-using-large-language-models/", "bibtex_url": "https://ylab.top/bib/scalable-information-extraction-from-free-text-electronic-health-records-using-large-language-models.bib", "citation_key": "Gu2025Scalableinformation", "bibtex_type": "article", "doi": "10.1186/s12874-025-02470-z", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "25", "issue": "1", "pages": "23", "publisher": "Springer Science and Business Media LLC", "abstract": "Background A vast amount of potentially useful information such as description of patient symptoms, family, and social history is recorded as free-text notes in electronic health records (EHRs) but is difficult to reliably extract at scale, limiting their utility in research. This study aims to assess whether an \"out of the box\" implementation of open-source large language models (LLMs) without any fine-tuning can accurately extract social determinants of health (SDoH) data from free-text clinical notes. Methods We conducted a cross-sectional study using EHR data from the Mass General Brigham (MGB) system, analyzing free-text notes for SDoH information. We selected a random sample of 200 patients and manually labeled nine SDoH aspects. Eight advanced open-source LLMs were evaluated against a baseline pattern-matching model. Two human reviewers provided the manual labels, achieving 93% inter-annotator agreement. LLM performance was assessed using accuracy metrics for overall, mentioned, and non-mentioned SDoH, and macro F1 scores. Results LLMs outperformed the baseline pattern-matching approach, particularly for explicitly mentioned SDoH, achieving up to 40% higher Accuracy mentioned . openchat_3.5 was the best-performing model, surpassing the baseline in overall accuracy across all nine SDoH aspects. The refined pipeline with prompt engineering reduced hallucinations and improved accuracy. Conclusions Open-source LLMs are effective and scalable tools for extracting SDoH from unstructured EHRs, surpassing traditional pattern-matching methods. Further refinement and domain-specific training could enhance their utility in clinical research and predictive analytics, improving healthcare outcomes and addressing health disparities.", "published_url": "https://doi.org/10.1186/s12874-025-02470-z", "crossref_type": "journal-article", "pmid": "39871166", "pmcid": "PMC11773977" }, { "title": "Analysis of Longitudinal Social Media for Monitoring Symptoms During a Pandemic", "authors": [ "Shixu Lin", "Lucas Garay", "Yining Hua", "Zhijiang Guo", "Wanxin Li", "Minghui Li", "Yujie Zhang", "Xiaolin Xu", "Jie Yang" ], "year": 2025, "venue": "Journal of Biomedical Informatics", "url": "https://www.sciencedirect.com/science/article/pii/S1532046424001825", "citation_page_url": "https://ylab.top/citation/analysis-of-longitudinal-social-media-for-monitoring-symptoms-during-a-pandemic/", "bibtex_url": "https://ylab.top/bib/analysis-of-longitudinal-social-media-for-monitoring-symptoms-during-a-pandemic.bib", "citation_key": "Lin2025Analysislongitudinal", "bibtex_type": "article", "doi": "10.1016/j.jbi.2025.104778", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "pmid": "39832606", "volume": "162", "pages": "104778", "publisher": "Elsevier BV", "abstract": "Objective Current studies leveraging social media data for disease monitoring face challenges like noisy colloquial language and insufficient tracking of user disease progression in longitudinal data settings. This study aims to develop a pipeline for collecting, cleaning, and analyzing large-scale longitudinal social media data for disease monitoring, with a focus on COVID-19 pandemic. Materials and methods This pipeline initiates by screening COVID-19 cases from tweets spanning February 1, 2020, to April 30, 2022. Longitudinal data is collected for each patient, two months before and three months after self-reporting. Symptoms are extracted using Name Entity Recognition (NER), followed by denoising with a combination of Graph Convolutional Network (GCN) and Bidirectional Encoder Representations from Transformers (BERT) model to retain only User-experienced Symptom Mentions (USM). Subsequently, symptoms are mapped to standardized medical concepts using the Unified Medical Language System (UMLS). Finally, this study conducts symptom pattern analysis and visualization to illustrate temporal changes in symptom prevalence and co-occurrence. Results This study identified 191,096 self-reported COVID-19-positive cases from COVID-19-related tweets and retrospectively collected 811,398,280 historical tweets, of which 2,120,964 contained symptoms information. After denoising, 39 % (832,287) of symptom-sharing tweets reflected user-experienced mentions. The trained USM model achieved an average F1 score of 0.927. Further analysis revealed a higher prevalence of upper respiratory tract symptoms during the Omicron period compared to the Delta and Wild-type periods. Additionally, there was a pronounced co-occurrence of lower respiratory tract and nervous system symptoms in the Wild-type strain and Delta variant. Conclusion This study established a robust framework for analyzing longitudinal social media data to monitor symptoms during a pandemic. By integrating denoising of user-experienced symptom mentions, our findings reveal the duration of different symptoms over time and by variant within a cohort of nearly 200,000 patients, providing critical insights into symptom trends that are often difficult to capture through traditional data source.", "published_url": "https://doi.org/10.1016/j.jbi.2025.104778", "crossref_type": "journal-article" }, { "title": "Identification of an ANCA-Associated Vasculitis Cohort Using Deep Learning and Electronic Health Records", "authors": [ "Liqin Wang", "John Novoa-Laurentiev", "Claire Cook", "Shruthi Srivatsan", "Yining Hua", "Jie Yang", "Eli Miloslavsky", "Hyon K Choi", "Li Zhou", "Zachary S Wallace" ], "year": 2025, "venue": "International Journal of Medical Informatics", "url": "https://www.sciencedirect.com/science/article/pii/S1386505624003551", "citation_page_url": "https://ylab.top/citation/identification-of-an-anca-associated-vasculitis-cohort-using-deep-learning-and-electronic-health-records/", "bibtex_url": "https://ylab.top/bib/identification-of-an-anca-associated-vasculitis-cohort-using-deep-learning-and-electronic-health-records.bib", "citation_key": "Wang2025Identificationanca", "bibtex_type": "article", "doi": "10.1016/j.ijmedinf.2025.105797", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "196", "pages": "105797", "publisher": "Elsevier BV", "abstract": "Background ANCA-associated vasculitis (AAV) is a rare but serious disease. Traditional case-identification methods using claims data can be time-intensive and may miss important subgroups. We hypothesized that a deep learning model analyzing electronic health records (EHR) can more accurately identify AAV cases. Methods We examined the Mass General Brigham (MGB) repository of clinical documentation from 12/1/1979 to 5/11/2021, using expert-curated keywords and ICD codes to identify a large cohort of potential AAV cases. Three labeled datasets (I, II, III) were created, each containing note sections. We trained and evaluated a range of machine learning and deep learning algorithms for note-level classification, using metrics like positive predictive value (PPV), sensitivity, F-score, area under the receiver operating characteristic curve (AUROC), and area under the precision and recall curve (AUPRC). The hierarchical attention network (HAN) was further evaluated for its ability to classify AAV cases at the patient-level, compared with rule-based algorithms in 2000 randomly chosen samples. Results Datasets I, II, and III comprised 6000, 3008, and 7500 note sections, respectively. HAN achieved the highest AUROC in all three datasets, with scores of 0.983, 0.991, and 0.991. The deep learning approach also had among the highest PPVs across the three datasets (0.941, 0.954, and 0.800, respectively). In a test cohort of 2000 cases, the HAN model achieved a PPV of 0.262 and an estimated sensitivity of 0.975. Compared to the best rule-based algorithm, HAN identified six additional AAV cases, representing 13% of the total. Conclusion The deep learning model effectively classifies clinical note sections for AAV diagnosis. Its application to EHR notes can potentially uncover additional cases missed by traditional rule-based methods.", "published_url": "https://doi.org/10.1016/j.ijmedinf.2025.105797", "crossref_type": "journal-article", "pmid": "39864108" }, { "title": "Characterizing Public Sentiments and Drug Interactions during COVID-19: A Pretrained Language Model and Network Analysis of Social Media Discourse", "authors": [ "Wanxin Li", "Yining Hua", "Peilin Zhou", "Li Zhou", "Xin Xu", "Jie Yang" ], "year": 2025, "venue": "Journal of Medical Internet Research", "url": "https://www.jmir.org/2025/1/e63755/", "citation_page_url": "https://ylab.top/citation/characterizing-public-sentiments-and-drug-interactions-during-covid-19-a-pretrained-language-model-and-network-analysis-of-social-media-discourse/", "bibtex_url": "https://ylab.top/bib/characterizing-public-sentiments-and-drug-interactions-during-covid-19-a-pretrained-language-model-and-network-analysis-of-social-media-discourse.bib", "citation_key": "Li2025Characterizingpublic", "bibtex_type": "article", "doi": "10.2196/63755", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "preprint_url": "https://doi.org/10.1101/2024.06.06.24308537", "publisher": "openRxiv", "abstract": "Objective Harnessing drug-related data posted on social media in real time can offer insights into how the pandemic impacts drug use and monitor misinformation. This study developed a natural language processing (NLP) pipeline tailored for the analysis of social media discourse on COVID-19 related drugs. Methods This study constructed a full pipeline for COVID-19 related drug tweet analysis, utilizing pre-trained language model-based NLP techniques as the backbone. This pipeline is architecturally composed of four core modules: named entity recognition (NER) and normalization to identify medical entities from relevant tweets and standardize them to uniform medication names, target sentiment analysis (TSA) to reveal sentiment polarities associated with the entities, topic modeling to understand underlying themes discussed by the population, and drug network analysis to potential adverse drug reactions (ADR) and drug-drug interactions (DDI). The pipeline was deployed to analyze tweets related to COVID-19 and drug therapies between February 1, 2020, and April 30, 2022. Results From a dataset comprising 2,124,757 relevant tweets sourced from 1,800,372 unique users, our NER model identified the top five most-discussed drugs: Ivermectin, Hydroxychloroquine, Remdesivir, Zinc, and Vitamin D. Sentiment and topic analysis revealed that public perception was predominantly shaped by celebrity endorsements, media hotspots, and governmental directives rather than empirical evidence of drug efficacy. Co-occurrence matrices and complex network analysis further identified emerging patterns of DDI and ADR that could be critical for public health surveillance like better safeguarding public safety in medicines use. Conclusion This study evidences that an NLP-based pipeline can be a robust tool for large-scale public health monitoring and can offer valuable supplementary data for traditional epidemiological studies concerning DDI and ADR. The framework presented here aspires to serve as a cornerstone for future social media-based public health analytics.", "published_url": "https://doi.org/10.1101/2024.06.06.24308537", "published_year": 2024, "crossref_type": "posted-content", "preprint_doi": "10.1101/2024.06.06.24308537" }, { "title": "Probabilistic Medical Predictions of Large Language Models", "authors": [ "Bowen Gu", "Rishi J. Desai", "Kueiyu Joshua Lin", "Jie Yang" ], "year": 2024, "venue": "npj Digital Medicine", "url": "https://www.nature.com/articles/s41746-024-01366-4", "citation_page_url": "https://ylab.top/citation/probabilistic-medical-predictions-of-large-language-models/", "bibtex_url": "https://ylab.top/bib/probabilistic-medical-predictions-of-large-language-models.bib", "citation_key": "Gu2024Probabilisticmedical", "bibtex_type": "article", "doi": "10.1038/s41746-024-01366-4", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "pmid": "39702641", "pmcid": "PMC11659327", "volume": "7", "issue": "1", "pages": "367", "publisher": "Springer Science and Business Media LLC", "abstract": "Large Language Models (LLMs) have shown promise in clinical applications through prompt engineering, allowing flexible clinical predictions. However, they struggle to produce reliable prediction probabilities, which are crucial for transparency and decision-making. While explicit prompts can lead LLMs to generate probability estimates, their numerical reasoning limitations raise concerns about reliability. We compared explicit probabilities from text generation to implicit probabilities derived from the likelihood of predicting the correct label token. Across six advanced open-source LLMs and five medical datasets, explicit probabilities consistently underperformed implicit probabilities in discrimination, precision, and recall. This discrepancy is more pronounced with smaller LLMs and imbalanced datasets, highlighting the need for cautious interpretation, improved probability estimation methods, and further research for clinical use of LLMs.", "published_url": "https://doi.org/10.1038/s41746-024-01366-4", "crossref_type": "journal-article" }, { "title": "Enhancing Postmarketing Surveillance of Medical Products with Large Language Models", "authors": [ "Michael E Matheny", "Jie Yang", "Joshua C Smith", "Colin G Walsh", "Mohammed A Al-Garadi", "Sharon E Davis", "Keith Marsolo", "Daniel Fabbri", "Ruth Reeves", "Kevin Johnson", "Gerald Dal Pan", "Robert Ball", "Rishi Desai" ], "year": 2024, "venue": "JAMA Network Open", "url": "https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2822418", "citation_page_url": "https://ylab.top/citation/enhancing-postmarketing-surveillance-of-medical-products-with-large-language-models/", "bibtex_url": "https://ylab.top/bib/enhancing-postmarketing-surveillance-of-medical-products-with-large-language-models.bib", "citation_key": "Matheny2024Enhancingpostmarketing", "bibtex_type": "article", "doi": "10.1001/jamanetworkopen.2024.28276", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "7", "issue": "8", "pages": "e2428276", "publisher": "American Medical Association (AMA)", "abstract": "Importance The Sentinel System is a key component of the US Food and Drug Administration (FDA) postmarketing safety surveillance commitment and uses clinical health care data to conduct analyses to inform drug labeling and safety communications, FDA advisory committee meetings, and other regulatory decisions. However, observational data are frequently deemed insufficient for reliable evaluation of safety concerns owing to limitations in underlying data or methodology. Advances in large language models (LLMs) provide new opportunities to address some of these limitations. However, careful consideration is necessary for how and where LLMs can be effectively deployed for these purposes. Observations LLMs may provide new avenues to support signal-identification activities to identify novel adverse event signals from narrative text of electronic health records. These algorithms may be used to support epidemiologic investigations examining the causal relationship between exposure to a medical product and an adverse event through development of probabilistic phenotyping of health outcomes of interest and extraction of information related to important confounding factors. LLMs may perform like traditional natural language processing tools by annotating text with controlled vocabularies with additional tailored training activities. LLMs offer opportunities for enhancing information extraction from adverse event reports, medical literature, and other biomedical knowledge sources. There are several challenges that must be considered when leveraging LLMs for postmarket surveillance. Prompt engineering is needed to ensure that LLM-extracted associations are accurate and specific. LLMs require extensive infrastructure to use, which many health care systems lack, and this can impact diversity, equity, and inclusion, and result in obscuring significant adverse event patterns in some populations. LLMs are known to generate nonfactual statements, which could lead to false positive signals and downstream evaluation activities by the FDA and other entities, incurring substantial cost. Conclusions and relevance LLMs represent a novel paradigm that may facilitate generation of information to support medical product postmarket surveillance activities that have not been possible. However, additional work is required to ensure LLMs can be used in a fair and equitable manner, minimize false positive findings, and support the necessary rigor of signal detection needed for regulatory activities.", "published_url": "https://doi.org/10.1001/jamanetworkopen.2024.28276", "crossref_type": "journal-article", "pmid": "39150707" }, { "title": "Clinical Text Datasets for Medical Artificial Intelligence and Large Language Models — A Systematic Review", "authors": [ "Jiageng Wu", "Xiaocong Liu", "Minghui Li", "Wanxin Li", "Zichang Su", "Lucas Garay", "Zhiyun Zhang", "Yujie Zhang", "Qingcheng Zeng", "Jie Shen", "Changzheng Yuan", "Jie Yang" ], "year": 2024, "venue": "New England Journal of Medicine AI", "url": "https://ai.nejm.org/doi/full/10.1056/AIra2400012", "citation_page_url": "https://ylab.top/citation/clinical-text-datasets-for-medical-artificial-intelligence-and-large-language-models-a-systematic-review/", "bibtex_url": "https://ylab.top/bib/clinical-text-datasets-for-medical-artificial-intelligence-and-large-language-models-a-systematic-review.bib", "citation_key": "Wu2024Clinicaltext", "bibtex_type": "article", "doi": "10.1056/aira2400012", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "volume": "1", "issue": "6", "publisher": "Massachusetts Medical Society", "published_url": "https://doi.org/10.1056/aira2400012", "crossref_type": "journal-article" }, { "title": "Streamlining Social Media Information Retrieval for Public Health Research with Deep Learning", "authors": [ "Yining Hua", "Jiageng Wu", "Shixu Lin", "Minghui Li", "Yujie Zhang", "Dinah Foer", "Siwen Wang", "Peilin Zhou", "Jie Yang", "Li Zhou" ], "year": 2024, "venue": "Journal of the American Medical Informatics Association", "url": "https://academic.oup.com/jamia/article/31/7/1569/7667306", "citation_page_url": "https://ylab.top/citation/streamlining-social-media-information-retrieval-for-public-health-research-with-deep-learning/", "bibtex_url": "https://ylab.top/bib/streamlining-social-media-information-retrieval-for-public-health-research-with-deep-learning.bib", "citation_key": "Hua2024Streamliningsocial", "bibtex_type": "article", "doi": "10.1093/jamia/ocae118", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "31", "issue": "7", "pages": "1569-1577", "publisher": "Oxford University Press (OUP)", "abstract": "Objective Social media-based public health research is crucial for epidemic surveillance, but most studies identify relevant corpora with keyword-matching. This study develops a system to streamline the process of curating colloquial medical dictionaries. We demonstrate the pipeline by curating a Unified Medical Language System (UMLS)-colloquial symptom dictionary from COVID-19-related tweets as proof of concept. Methods COVID-19-related tweets from February 1, 2020, to April 30, 2022 were used. The pipeline includes three modules: a named entity recognition module to detect symptoms in tweets; an entity normalization module to aggregate detected entities; and a mapping module that iteratively maps entities to Unified Medical Language System concepts. A random 500 entity samples were drawn from the final dictionary for accuracy validation. Additionally, we conducted a symptom frequency distribution analysis to compare our dictionary to a pre-defined lexicon from previous research. Results We identified 498 480 unique symptom entity expressions from the tweets. Pre-processing reduces the number to 18 226. The final dictionary contains 38 175 unique expressions of symptoms that can be mapped to 966 UMLS concepts (accuracy = 95%). Symptom distribution analysis found that our dictionary detects more symptoms and is effective at identifying psychiatric disorders like anxiety and depression, often missed by pre-defined lexicons. Conclusions This study advances public health research by implementing a novel, systematic pipeline for curating symptom lexicons from social media data. The final lexicon's high accuracy, validated by medical professionals, underscores the potential of this methodology to reliably interpret, and categorize vast amounts of unstructured social media data into actionable medical insights across diverse linguistic and regional landscapes.", "published_url": "https://doi.org/10.1093/jamia/ocae118", "crossref_type": "journal-article", "pmid": "38718216", "pmcid": "PMC11187427" }, { "title": "Guiding Clinical Reasoning with Large Language Models via Knowledge Seeds", "authors": [ "Jiageng Wu", "Xian Wu", "Jie Yang" ], "year": 2024, "venue": "International Joint Conference on Artificial Intelligence (IJCAI)", "url": "https://arxiv.org/abs/2403.06609", "citation_page_url": "https://ylab.top/citation/guiding-clinical-reasoning-with-large-language-models-via-knowledge-seeds/", "bibtex_url": "https://ylab.top/bib/guiding-clinical-reasoning-with-large-language-models-via-knowledge-seeds.bib", "citation_key": "Wu2024Guidingclinical", "bibtex_type": "inproceedings", "doi": "10.24963/ijcai.2024/829", "metadata_sources": [ "Crossref" ], "metadata_match": { "crossref": 1.0 }, "preprint_url": "https://arxiv.org/abs/2403.06609", "pages": "7491-7499", "publisher": "International Joint Conferences on Artificial Intelligence Organization", "abstract": "Clinical reasoning refers to the cognitive process that physicians employ in evaluating and managing patients. This process typically involves suggesting necessary examinations, diagnosing patients’ diseases, and selecting appropriate therapies, etc. Accurate clinical reasoning requires extensive medical knowledge and rich clinical experience, setting a high bar for physicians. This is particularly challenging in developing countries due to the overwhelming number of patients and limited physician resources, contributing significantly to global health inequity and necessitating automated clinical reasoning approaches. Recently, the emergence of large language models (LLMs) such as ChatGPT and GPT-4 have demonstrated their potential in clinical reasoning. However, these LLMs are prone to hallucination problems, and the reasoning process of LLMs may not align with the clinical decision pathways of physicians. In this study, we introduce a novel framework, In-Context Padding (ICP), to enhance LLMs reasoning with medical knowledge. Specifically, we infer critical clinical reasoning elements (referred to as knowledge seeds) and use these as anchors to guide the generation process of LLMs. Experiments on two clinical question datasets validate that ICP significantly improves the clinical reasoning ability of LLMs.", "published_url": "https://doi.org/10.24963/ijcai.2024/829", "crossref_type": "proceedings-article", "arxiv_id": "2403.06609" }, { "title": "Large Language Models Leverage External Knowledge to Extend Clinical Insight Beyond Language Boundaries", "authors": [ "Jiageng Wu", "Xian Wu", "Zhaopeng Qiu", "Minghui Li", "Shixu Lin", "Yingying Zhang", "Yefeng Zheng", "Changzheng Yuan", "Jie Yang" ], "year": 2024, "venue": "Journal of the American Medical Informatics Association", "url": "https://academic.oup.com/jamia/article/31/6/1394/7659846", "citation_page_url": "https://ylab.top/citation/large-language-models-leverage-external-knowledge-to-extend-clinical-insight-beyond-language-boundaries/", "bibtex_url": "https://ylab.top/bib/large-language-models-leverage-external-knowledge-to-extend-clinical-insight-beyond-language-boundaries.bib", "citation_key": "Wu2024Largelanguage", "bibtex_type": "article", "doi": "10.1093/jamia/ocae079", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "31", "issue": "9", "pages": "2054-2064", "publisher": "Oxford University Press (OUP)", "abstract": "Objectives Large Language Models (LLMs) such as ChatGPT and Med-PaLM have excelled in various medical question-answering tasks. However, these English-centric models encounter challenges in non-English clinical settings, primarily due to limited clinical knowledge in respective languages, a consequence of imbalanced training corpora. We systematically evaluate LLMs in the Chinese medical context and develop a novel in-context learning framework to enhance their performance. Materials and methods The latest China National Medical Licensing Examination (CNMLE-2022) served as the benchmark. We collected 53 medical books and 381 149 medical questions to construct the medical knowledge base and question bank. The proposed Knowledge and Few-shot Enhancement In-context Learning (KFE) framework leverages the in-context learning ability of LLMs to integrate diverse external clinical knowledge sources. We evaluated KFE with ChatGPT (GPT-3.5), GPT-4, Baichuan2-7B, Baichuan2-13B, and QWEN-72B in CNMLE-2022 and further investigated the effectiveness of different pathways for incorporating LLMs with medical knowledge from 7 distinct perspectives. Results Directly applying ChatGPT failed to qualify for the CNMLE-2022 at a score of 51. Cooperated with the KFE framework, the LLMs with varying sizes yielded consistent and significant improvements. The ChatGPT's performance surged to 70.04 and GPT-4 achieved the highest score of 82.59. This surpasses the qualification threshold (60) and exceeds the average human score of 68.70, affirming the effectiveness and robustness of the framework. It also enabled a smaller Baichuan2-13B to pass the examination, showcasing the great potential in low-resource settings. Discussion and conclusion This study shed light on the optimal practices to enhance the capabilities of LLMs in non-English medical scenarios. By synergizing medical knowledge through in-context learning, LLMs can extend clinical insight beyond language barriers in healthcare, significantly reducing language-related disparities of LLM applications and ensuring global benefit in this field.", "published_url": "https://doi.org/10.1093/jamia/ocae079", "crossref_type": "journal-article", "pmid": "38684792", "pmcid": "PMC11339525" }, { "title": "Revealing COVID-19's Social Dynamics: Diachronic Semantic Analysis of Vaccine and Symptom Discourse on Twitter", "authors": [ "Zeqiang Wang", "Jiageng Wu", "Yuqi Wang", "Wei Wang", "Jie Yang", "Nishanth R Sastry", "Jon Johnson", "Suparna De" ], "year": 2024, "venue": "Conference on Empirical Methods in Natural Language Processing (EMNLP), Findings", "url": null, "citation_page_url": "https://ylab.top/citation/revealing-covid-19-s-social-dynamics-diachronic-semantic-analysis-of-vaccine-and-symptom-discourse-on-twitter/", "bibtex_url": "https://ylab.top/bib/revealing-covid-19-s-social-dynamics-diachronic-semantic-analysis-of-vaccine-and-symptom-discourse-on-twitter.bib", "citation_key": "Wang2024Revealingcovid", "bibtex_type": "inproceedings", "doi": "10.18653/v1/2024.findings-emnlp.193", "metadata_sources": [ "Crossref" ], "metadata_match": { "crossref": 0.9954 }, "pages": "3383-3394", "publisher": "Association for Computational Linguistics", "published_url": "https://doi.org/10.18653/v1/2024.findings-emnlp.193", "published_title": "Revealing COVID-19’s Social Dynamics: Diachronic Semantic Analysis of Vaccine and Symptom Discourse on Twitter", "crossref_type": "proceedings-article" }, { "title": "MedJourney: Benchmark and Evaluation of Large Language Models over Patient Clinical Journey", "authors": [ "Xian Wu", "Yutian Zhao", "Yunyan Zhang", "Jiageng Wu", "Zhihong Zhu", "Yingying Zhang", "Yi Ouyang", "Ziheng Zhang", "Huimin Wang", "Zhenxi Lin", "Jie Yang" ], "year": 2024, "venue": "Conference on Neural Information Processing Systems (NeurIPS)", "url": "https://openreview.net/forum?id=LJ4sBAFoNX", "citation_page_url": "https://ylab.top/citation/medjourney-benchmark-and-evaluation-of-large-language-models-over-patient-clinical-journey/", "bibtex_url": "https://ylab.top/bib/medjourney-benchmark-and-evaluation-of-large-language-models-over-patient-clinical-journey.bib", "citation_key": "Wu2024Medjourneybenchmark", "bibtex_type": "inproceedings", "doi": "10.52202/079017-2782", "metadata_sources": [ "Crossref" ], "metadata_match": { "crossref": 1.0 }, "code_url": "https://github.com/Medical-AI-Learning/MedJourney", "data_url": "https://github.com/Medical-AI-Learning/MedJourney", "pages": "87621-87646", "publisher": "Neural Information Processing Systems Foundation, Inc. (NeurIPS)", "published_url": "https://doi.org/10.52202/079017-2782", "crossref_type": "proceedings-article" }, { "title": "The Sexual and Gender-Diverse Face More Health Challenges During COVID-19: A Large-Scale Social Media Analysis with Natural Language Processing", "authors": [ "Zhiyun Zhang", "Yining Hua", "Peilin Zhou", "Shixu Lin", "Minghui Li", "Yujie Zhang", "Li Zhou", "Yanhui Liao", "Jie Yang" ], "year": 2024, "venue": "Health Data Science", "url": "https://spj.science.org/doi/10.34133/hds.0127", "citation_page_url": "https://ylab.top/citation/the-sexual-and-gender-diverse-face-more-health-challenges-during-covid-19-a-large-scale-social-media-analysis-with-natural-language-processing/", "bibtex_url": "https://ylab.top/bib/the-sexual-and-gender-diverse-face-more-health-challenges-during-covid-19-a-large-scale-social-media-analysis-with-natural-language-processing.bib", "citation_key": "Zhang2024Sexualgender", "bibtex_type": "article", "doi": "10.34133/hds.0127", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 0.9452, "crossref": 1.0, "europe_pmc": 1.0 }, "preprint_url": "https://doi.org/10.34133/hds.0127", "pmid": "39247070", "pmcid": "PMC11378377", "publisher": "openRxiv", "abstract": "Background: The COVID-19 pandemic has caused a disproportionate impact on the sex and gender diversity (SGD) community. Compared with non-SGD populations, their social relations and health status are more vulnerable, whereas public health data regarding SGD is scarce. Methods: : To analyze the concerns and health status of SGD individuals, this cohort study leveraged 471,371,477 tweets from 251,455 SGD and 22,644,411 non-SGD users, spanning from February 1, 2020, to April 30, 2022. The outcome measures comprised the distribution and dynamics of COVID-related topics, attitudes towards vaccines and the prevalence of symptoms. Results: : Topic analysis revealed that SGD users engaged more frequently in discussions related to “friends and family” (20.5% vs 13.1%, P Conclusion: The results provide insights to an understanding of the unique needs and experiences of the SGD community during the pandemic, emphasizing the value of social media data in epidemiological and public health research.", "published_url": "https://doi.org/10.1101/2024.06.14.24308944", "crossref_type": "posted-content", "preprint_doi": "10.34133/hds.0127" }, { "title": "Uncovering Language Disparity of ChatGPT on Retinal Vascular Disease Classification: Cross-Sectional Study", "authors": [ "Xiaocong Liu", "Jiageng Wu", "An Shao", "Wenyue Shen", "Panpan Ye", "Yao Wang", "Juan Ye", "Kai Jin", "Jie Yang" ], "year": 2024, "venue": "Journal of Medical Internet Research", "url": "https://www.jmir.org/2024/1/e51926", "citation_page_url": "https://ylab.top/citation/uncovering-language-disparity-of-chatgpt-on-retinal-vascular-disease-classification-cross-sectional-study/", "bibtex_url": "https://ylab.top/bib/uncovering-language-disparity-of-chatgpt-on-retinal-vascular-disease-classification-cross-sectional-study.bib", "citation_key": "Liu2024Uncoveringlanguage", "bibtex_type": "article", "doi": "10.2196/51926", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "26", "pages": "e51926", "publisher": "JMIR Publications Inc.", "abstract": "Background Benefiting from rich knowledge and the exceptional ability to understand text, large language models like ChatGPT have shown great potential in English clinical environments. However, the performance of ChatGPT in non-English clinical settings, as well as its reasoning, have not been explored in depth. Objective This study aimed to evaluate ChatGPT's diagnostic performance and inference abilities for retinal vascular diseases in a non-English clinical environment. Methods In this cross-sectional study, we collected 1226 fundus fluorescein angiography reports and corresponding diagnoses written in Chinese and tested ChatGPT with 4 prompting strategies (direct diagnosis or diagnosis with a step-by-step reasoning process and in Chinese or English). Results Compared with ChatGPT using Chinese prompts for direct diagnosis that achieved an F 1 -score of 70.47%, ChatGPT using English prompts for direct diagnosis achieved the best diagnostic performance (80.05%), which was inferior to ophthalmologists (89.35%) but close to ophthalmologist interns (82.69%). As for its inference abilities, although ChatGPT can derive a reasoning process with a low error rate (0.4 per report) for both Chinese and English prompts, ophthalmologists identified that the latter brought more reasoning steps with less incompleteness (44.31%), misinformation (1.96%), and hallucinations (0.59%) (all P 1 -score (P=.04) between Chinese and English prompts. In short, when prompted in English, ChatGPT exhibited enhanced diagnostic and inference capabilities for retinal vascular disease classification based on Chinese fundus fluorescein angiography reports. Conclusions ChatGPT can serve as a helpful medical assistant to provide diagnosis in non-English clinical environments, but there are still performance gaps, language disparities, and errors compared to professionals, which demonstrate the potential limitations and the need to continually explore more robust large language models in ophthalmology practice.", "published_url": "https://doi.org/10.2196/51926", "crossref_type": "journal-article", "pmid": "38252483", "pmcid": "PMC10845019" }, { "title": "GreenPLM: Cross-Lingual Transfer of Monolingual Pre-Trained Language Models at Almost No Cost", "authors": [ "Qingcheng Zeng", "Lucas Garay", "Peilin Zhou", "Dading Chong", "Yining Hua", "Jiageng Wu", "Yikang Pan", "Han Zhou", "Rob Voigt", "Jie Yang" ], "year": 2023, "venue": "International Joint Conference on Artificial Intelligence (IJCAI)", "url": "https://arxiv.org/abs/2211.06993", "citation_page_url": "https://ylab.top/citation/greenplm-cross-lingual-transfer-of-monolingual-pre-trained-language-models-at-almost-no-cost/", "bibtex_url": "https://ylab.top/bib/greenplm-cross-lingual-transfer-of-monolingual-pre-trained-language-models-at-almost-no-cost.bib", "citation_key": "Zeng2023Greenplmcross", "bibtex_type": "inproceedings", "doi": "10.24963/ijcai.2023/698", "metadata_sources": [ "Crossref" ], "metadata_match": { "crossref": 1.0 }, "code_url": "https://github.com/qcznlp/GreenPLMs", "preprint_url": "https://arxiv.org/abs/2211.06993", "pages": "6290-6298", "publisher": "International Joint Conferences on Artificial Intelligence Organization", "abstract": "Large pre-trained models have revolutionized natural language processing (NLP) research and applications, but high training costs and limited data resources have prevented their benefits from being shared equally amongst speakers of all the world's languages. To address issues of cross-linguistic access to such models and reduce energy consumption for sustainability during large-scale model training, this study proposes an effective and energy-efficient framework called GreenPLM that uses bilingual lexicons to directly ``translate'' pre-trained language models of one language into another at almost no additional cost. We validate this approach in 18 languages' BERT models and show that this framework is comparable to, if not better than, other heuristics with high training costs. In addition, given lightweight continued pre-training on limited data where available, this framework outperforms the original monolingual language models in six out of seven tested languages with up to 200x less pre-training efforts. Aiming at the Leave No One Behind Principle (LNOB), our approach manages to reduce inequalities between languages and energy consumption greatly. We make our codes and models publicly available at https://github.com/qcznlp/GreenPLMs.", "published_url": "https://doi.org/10.24963/ijcai.2023/698", "crossref_type": "proceedings-article", "arxiv_id": "2211.06993" }, { "title": "Efficacy of WeChat-Based Online Smoking Cessation Intervention ('WeChat WeQuit') in China: A Randomised Controlled Trial", "authors": [ "Jinsong Tang", "Jie Yang", "Yi Liu", "Xiaocong Liu", "Ling Li", "Yunkai Sun", "Jieyin Jin", "Yehong Fang", "Zitang Zhou", "Yunfei Wang", "Yueheng Liu", "Wei Chen", "Ann McNeill", "Brian Kelly", "Joanna E Cohen", "Yanhui Liao" ], "year": 2023, "venue": "eClinicalMedicine", "url": "https://www.thelancet.com/journals/eclinm/article/PIIS2589-5370(23)00186-4/fulltext", "citation_page_url": "https://ylab.top/citation/efficacy-of-wechat-based-online-smoking-cessation-intervention-wechat-wequit-in-china-a-randomised-controlled-trial/", "bibtex_url": "https://ylab.top/bib/efficacy-of-wechat-based-online-smoking-cessation-intervention-wechat-wequit-in-china-a-randomised-controlled-trial.bib", "citation_key": "Tang2023Efficacywechat", "bibtex_type": "article", "doi": "10.1016/j.eclinm.2023.102009", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "60", "pages": "102009", "publisher": "Elsevier BV", "abstract": "Background China has approximately 300 million current smokers, and smoking cessation services are limited. This study aimed to assess the efficacy of a Cognitive Behavioral Theory-based smoking cessation intervention ('WeChat WeQuit') via the most popular social media platform in China, WeChat. Methods A parallel, single-blind, two-arm randomised controlled trial was conducted via WeChat between March 19, 2020 and November 16, 2022. Chinese-speaking adult smokers (n = 2000) willing to quit within one month were recruited and randomised in a 1:1 ratio. The intervention group (n = 1005) received the 'WeChat WeQuit' program and the control group (n = 955) received control messages for 14 weeks (2-week prequit and 12-week postquit). Participants were followed up to 26 weeks after the quit date. The primary outcome was self-reported continuous smoking abstinence rate, biochemically validated at 26 weeks. The secondary outcomes were self-reported 7-day and continuous abstinence rates at 6 months. All analyses were by intention to treat. The trial is registered at ClinicalTrials.gov (NCT03169686). Findings By intention-to-treat analysis, the biochemically verified 26-week continuous abstinence rate was 11.94% in the intervention group and 2.81% in the control group (OR = 4.68, 95% CI: 3.07-7.13, p p Interpretation The 'WeChat WeQuit' intervention significantly increased smoking abstinence rates at 6 months and should be considered for treatment-seeking smokers in China. Funding The research is supported by the Natural Science Foundation of Hunan Province (2020JJ4794, YLiao), the K.C. Wong Postdoctoral Fellowship for YLiao to study at King's College London, and China Medical Board (CMB) Open Competition Program (grant no. 15-226, 22-485, YLiao).", "published_url": "https://doi.org/10.1016/j.eclinm.2023.102009", "crossref_type": "journal-article", "pmid": "37251625", "pmcid": "PMC10209122" }, { "title": "Exploring Social Media for Early Detection of Depression in COVID-19 Patients", "authors": [ "Jiageng Wu", "Xian Wu", "Yining Hua", "Shixu Lin", "Yefeng Zheng", "Jie Yang" ], "year": 2023, "venue": "The Web Conference (WWW)", "url": "https://arxiv.org/abs/2302.12044", "citation_page_url": "https://ylab.top/citation/exploring-social-media-for-early-detection-of-depression-in-covid-19-patients/", "bibtex_url": "https://ylab.top/bib/exploring-social-media-for-early-detection-of-depression-in-covid-19-patients.bib", "citation_key": "Wu2023Exploringsocial", "bibtex_type": "inproceedings", "doi": "10.1145/3543507.3583867", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "preprint_url": "https://arxiv.org/abs/2302.12044", "pages": "3968-3977", "publisher": "ACM", "abstract": "The COVID-19 pandemic has caused substantial damage to global health. Even though three years have passed, the world continues to struggle with the virus. Concerns are growing about the impact of COVID-19 on the mental health of infected individuals, who are more likely to experience depression, which can have long-lasting consequences for both the affected individuals and the world. Detection and intervention at an early stage can reduce the risk of depression in COVID-19 patients. In this paper, we investigated the relationship between COVID-19 infection and depression through social media analysis. Firstly, we managed a dataset of COVID-19 patients that contains information about their social media activity both before and after infection. Secondly,We conducted an extensive analysis of this dataset to investigate the characteristic of COVID-19 patients with a higher risk of depression. Thirdly, we proposed a deep neural network for early prediction of depression risk. This model considers daily mood swings as a psychiatric signal and incorporates textual and emotional characteristics via knowledge distillation. Experimental results demonstrate that our proposed framework outperforms baselines in detecting depression risk, with an AUROC of 0.9317 and an AUPRC of 0.8116. Our model has the potential to enable public health organizations to initiate prompt intervention with high-risk patients", "published_url": "https://doi.org/10.1145/3543507.3583867", "crossref_type": "proceedings-article", "arxiv_id": "2302.12044" }, { "title": "Trend and Co-occurrence Network of COVID-19 Symptoms Through Large-Scale Social Media: Infoveillance Study", "authors": [ "Jiageng Wu", "Lumin Wang", "Yining Hua", "Minghui Li", "Li Zhou", "David W Bates", "Jie Yang" ], "year": 2023, "venue": "Journal of Medical Internet Research", "url": "https://www.jmir.org/2023/1/e45419", "citation_page_url": "https://ylab.top/citation/trend-and-co-occurrence-network-of-covid-19-symptoms-through-large-scale-social-media-infoveillance-study/", "bibtex_url": "https://ylab.top/bib/trend-and-co-occurrence-network-of-covid-19-symptoms-through-large-scale-social-media-infoveillance-study.bib", "citation_key": "Wu2023Trendco", "bibtex_type": "article", "doi": "10.2196/45419", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 0.9434, "crossref": 0.9434 }, "volume": "25", "pages": "e45419", "publisher": "JMIR Publications Inc.", "abstract": "Background For an emergent pandemic, such as COVID-19, the statistics of symptoms based on hospital data may be biased or delayed due to the high proportion of asymptomatic or mild-symptom infections that are not recorded in hospitals. Meanwhile, the difficulty in accessing large-scale clinical data also limits many researchers from conducting timely research. Objective Given the wide coverage and promptness of social media, this study aimed to present an efficient workflow to track and visualize the dynamic characteristics and co-occurrence of symptoms for the COVID-19 pandemic from large-scale and long-term social media data. Methods This retrospective study included 471,553,966 COVID-19–related tweets from February 1, 2020, to April 30, 2022. We curated a hierarchical symptom lexicon for social media containing 10 affected organs/systems, 257 symptoms, and 1808 synonyms. The dynamic characteristics of COVID-19 symptoms over time were analyzed from the perspectives of weekly new cases, overall distribution, and temporal prevalence of reported symptoms. The symptom evolutions between virus strains (Delta and Omicron) were investigated by comparing the symptom prevalence during their dominant periods. A co-occurrence symptom network was developed and visualized to investigate inner relationships among symptoms and affected body systems. Results This study identified 201 COVID-19 symptoms and grouped them into 10 affected body systems. There was a significant correlation between the weekly quantity of self-reported symptoms and new COVID-19 infections (Pearson correlation coefficient=0.8528; P<.001). We also observed a 1-week leading trend (Pearson correlation coefficient=0.8802; P<.001) between them. The frequency of symptoms showed dynamic changes as the pandemic progressed, from typical respiratory symptoms in the early stage to more musculoskeletal and nervous symptoms in the later stages. We identified the difference in symptoms between the Delta and Omicron periods. There were fewer severe symptoms (coma and dyspnea), more flu-like symptoms (throat pain and nasal congestion), and fewer typical COVID symptoms (anosmia and taste altered) in the Omicron period than in the Delta period (all P<.001). Network analysis revealed co-occurrences among symptoms and systems corresponding to specific disease progressions, including palpitations (cardiovascular) and dyspnea (respiratory), and alopecia (musculoskeletal) and impotence (reproductive). Conclusions This study identified more and milder COVID-19 symptoms than clinical research and characterized the dynamic symptom evolution based on 400 million tweets over 27 months. The symptom network revealed potential comorbidity risk and prognostic disease progression. These findings demonstrate that the cooperation of social media and a well-designed workflow can depict a holistic picture of pandemic symptoms to complement clinical studies.", "published_url": "https://doi.org/10.2196/45419", "published_title": "Trend and Co-occurrence Network of COVID-19 Symptoms From Large-Scale Social Media Data: Infoveillance Study", "crossref_type": "journal-article" }, { "title": "Better Pay Attention Whilst Fuzzing", "authors": [ "Shunkai Zhu", "Jingyi Wang", "Jun Sun", "Jie Yang", "Xingwei Lin", "Liyi Zhang", "Peng Cheng" ], "year": 2023, "venue": "IEEE Transactions on Software Engineering", "url": null, "citation_page_url": "https://ylab.top/citation/better-pay-attention-whilst-fuzzing/", "bibtex_url": "https://ylab.top/bib/better-pay-attention-whilst-fuzzing.bib", "citation_key": "Zhu2023Betterpay", "bibtex_type": "article", "doi": "10.1109/tse.2023.3338129", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "volume": "50", "issue": "2", "pages": "190-208", "publisher": "Institute of Electrical and Electronics Engineers (IEEE)", "published_url": "https://doi.org/10.1109/tse.2023.3338129", "published_year": 2024, "crossref_type": "journal-article" }, { "title": "Transforming Retinal Vascular Disease Classification: A Comprehensive Analysis of ChatGPT's Performance and Inference Abilities", "authors": [ "Xiaocong Liu", "Jiageng Wu", "An Shao", "Wenyue Shen", "Panpan Ye", "Yao Wang", "Juan Ye", "Kai Jin", "Jie Yang" ], "year": 2023, "venue": "American Medical Informatics Association Annual Symposium (AMIA)", "url": null, "citation_page_url": "https://ylab.top/citation/transforming-retinal-vascular-disease-classification-a-comprehensive-analysis-of-chatgpt-s-performance-and-inference-abilities/", "bibtex_url": "https://ylab.top/bib/transforming-retinal-vascular-disease-classification-a-comprehensive-analysis-of-chatgpt-s-performance-and-inference-abilities.bib", "citation_key": "Liu2023Transformingretinal", "bibtex_type": "inproceedings", "doi": null }, { "title": "YATO: Yet Another Deep Learning Based Text Analysis Open Toolkit", "authors": [ "Zeqiang Wang", "Yile Wang", "Jiageng Wu", "Zhiyang Teng", "Jie Yang" ], "year": 2023, "venue": "Conference on Empirical Methods in Natural Language Processing (EMNLP), Demo", "url": "https://arxiv.org/abs/2209.13877", "citation_page_url": "https://ylab.top/citation/yato-yet-another-deep-learning-based-text-analysis-open-toolkit/", "bibtex_url": "https://ylab.top/bib/yato-yet-another-deep-learning-based-text-analysis-open-toolkit.bib", "citation_key": "Wang2023Yatoyet", "bibtex_type": "inproceedings", "doi": "10.18653/v1/2023.emnlp-demo.11", "metadata_sources": [ "ORCID", "Crossref", "arXiv" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "code_url": "https://github.com/jiesutd/YATO", "preprint_url": "https://arxiv.org/abs/2209.13877", "publisher": "Association for Computational Linguistics", "abstract": "We introduce YATO, an open-source, easy-to-use toolkit for text analysis with deep learning. Different from existing heavily engineered toolkits and platforms, YATO is lightweight and user-friendly for researchers from cross-disciplinary areas. Designed in a hierarchical structure, YATO supports free combinations of three types of widely used features including 1) traditional neural networks (CNN, RNN, etc.); 2) pre-trained language models (BERT, RoBERTa, ELECTRA, etc.); and 3) user-customized neural features via a simple configurable file. Benefiting from the advantages of flexibility and ease of use, YATO can facilitate fast reproduction and refinement of state-of-the-art NLP models, and promote the cross-disciplinary applications of NLP techniques. The code, examples, and documentation are publicly available at https://github.com/jiesutd/YATO. A demo video is also available at https://www.youtube.com/playlist?list=PLJ0mhzMcRuDUlTkzBfAftOqiJRxYTTjXH.", "published_url": "https://doi.org/10.18653/v1/2023.emnlp-demo.11", "crossref_type": "proceedings-article", "arxiv_id": "2209.13877" }, { "title": "Genetically Predicted Sex Hormone Levels and Health Outcomes: Phenome-Wide Mendelian Randomization Investigation", "authors": [ "Shuai Yuan", "Lijuan Wang", "Jing Sun", "Lili Yu", "Xuan Zhou", "Jie Yang", "Yimin Zhu", "Dipender Gill", "Stephen Burgess", "Joshua C Denny" ], "year": 2022, "venue": "International Journal of Epidemiology", "url": "https://academic.oup.com/ije/article/51/6/1931/6590882", "citation_page_url": "https://ylab.top/citation/genetically-predicted-sex-hormone-levels-and-health-outcomes-phenome-wide-mendelian-randomization-investigation/", "bibtex_url": "https://ylab.top/bib/genetically-predicted-sex-hormone-levels-and-health-outcomes-phenome-wide-mendelian-randomization-investigation.bib", "citation_key": "Yuan2022Geneticallypredicted", "bibtex_type": "article", "doi": "10.1093/ije/dyac036", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "51", "issue": "6", "pages": "1931-1942", "publisher": "Oxford University Press (OUP)", "abstract": "Background Sex hormone-binding globulin (SHBG), testosterone and oestradiol have been associated with many diseases in observational studies; however, the causality of associations remains unestablished. Methods A phenome-wide Mendelian randomization (MR) association study was performed to explore disease outcomes associated with genetically proxied circulating SHBG, testosterone and oestradiol levels by using updated genetic instruments in 339 197 unrelated White British individuals (54% female) in the UK Biobank. Two-sample MR analyses with data from large genetic studies were conducted to replicate identified associations in phenome-wide MR analyses. Multivariable MR analyses were performed to investigate mediation effects of hormone-related biomarkers in observed associations with diseases. Results Phenome-wide MR analyses examined associations of genetically predicted SHBG, testosterone and oestradiol levels with 1211 disease outcomes, and identified 28 and 13 distinct phenotypes associated with genetically predicted SHBG and testosterone, respectively; 22 out of 28 associations for SHBG and 10 out of 13 associations for testosterone were replicated in two-sample MR analyses. Higher genetically predicted SHBG levels were associated with a reduced risk of hypertension, type 2 diabetes, diabetic complications, coronary atherosclerotic outcomes, gout and benign and malignant neoplasm of uterus, but an increased risk of varicose veins and fracture (mainly in females). Higher genetically predicted testosterone levels were associated with a lower risk of type 2 diabetes, coronary atherosclerotic outcomes, gout and coeliac disease mainly in males, but an increased risk of cholelithiasis in females. Conclusions These findings suggest that sex hormones may causally affect risk of several health outcomes.", "published_url": "https://doi.org/10.1093/ije/dyac036", "crossref_type": "journal-article", "pmid": "35218343", "pmcid": "PMC9749729" }, { "title": "METS-CoV: A Dataset of Medical Entity and Targeted Sentiment on COVID-19 Related Tweets", "authors": [ "Peilin Zhou", "Zeqiang Wang", "Dading Chong", "Zhijiang Guo", "Yining Hua", "Zichang Su", "Zhiyang Teng", "Jiageng Wu", "Jie Yang" ], "year": 2022, "venue": "Conference on Neural Information Processing Systems (NeurIPS)", "url": "https://openreview.net/forum?id=GP1Ncd8nTgn", "citation_page_url": "https://ylab.top/citation/mets-cov-a-dataset-of-medical-entity-and-targeted-sentiment-on-covid-19-related-tweets/", "bibtex_url": "https://ylab.top/bib/mets-cov-a-dataset-of-medical-entity-and-targeted-sentiment-on-covid-19-related-tweets.bib", "citation_key": "Zhou2022Metscov", "bibtex_type": "inproceedings", "doi": "10.52202/068431-1593", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "code_url": "https://github.com/YLab-Open/METS-CoV", "data_url": "https://github.com/YLab-Open/METS-CoV", "pages": "21916-21932", "publisher": "Neural Information Processing Systems Foundation, Inc. (NeurIPS)", "abstract": "The COVID-19 pandemic continues to bring up various topics discussed or debated on social media. In order to explore the impact of pandemics on people's lives, it is crucial to understand the public's concerns and attitudes towards pandemic-related entities (e.g., drugs, vaccines) on social media. However, models trained on existing named entity recognition (NER) or targeted sentiment analysis (TSA) datasets have limited ability to understand COVID-19-related social media texts because these datasets are not designed or annotated from a medical perspective. This paper releases METS-CoV, a dataset containing medical entities and targeted sentiments from COVID-19-related tweets. METS-CoV contains 10,000 tweets with 7 types of entities, including 4 medical entity types (Disease, Drug, Symptom, and Vaccine) and 3 general entity types (Person, Location, and Organization). To further investigate tweet users' attitudes toward specific entities, 4 types of entities (Person, Organization, Drug, and Vaccine) are selected and annotated with user sentiments, resulting in a targeted sentiment dataset with 9,101 entities (in 5,278 tweets). To the best of our knowledge, METS-CoV is the first dataset to collect medical entities and corresponding sentiments of COVID-19-related tweets. We benchmark the performance of classical machine learning models and state-of-the-art deep learning models on NER and TSA tasks with extensive experiments. Results show that the dataset has vast room for improvement for both NER and TSA tasks. METS-CoV is an important resource for developing better medical social media tools and facilitating computational social science research, especially in epidemiology. Our data, annotation guidelines, benchmark models, and source code are publicly available (https://github.com/YLab-Open/METS-CoV) to ensure reproducibility.", "published_url": "https://doi.org/10.52202/068431-1593", "crossref_type": "proceedings-article" }, { "title": "Tracking the Impact of COVID-19 and Lockdown Policies on Public Mental Health Using Social Media: Infoveillance Study", "authors": [ "Minghui Li", "Yining Hua", "Yanhui Liao", "Li Zhou", "Xue Li", "Ling Wang", "Jie Yang" ], "year": 2022, "venue": "Journal of Medical Internet Research", "url": "https://www.jmir.org/2022/10/e39676/", "citation_page_url": "https://ylab.top/citation/tracking-the-impact-of-covid-19-and-lockdown-policies-on-public-mental-health-using-social-media-infoveillance-study/", "bibtex_url": "https://ylab.top/bib/tracking-the-impact-of-covid-19-and-lockdown-policies-on-public-mental-health-using-social-media-infoveillance-study.bib", "citation_key": "Li2022Trackingimpact", "bibtex_type": "article", "doi": "10.2196/39676", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "24", "issue": "10", "pages": "e39676", "publisher": "JMIR Publications Inc.", "abstract": "Background The COVID-19 pandemic and its corresponding preventive and control measures have increased the mental burden on the public. Understanding and tracking changes in public mental status can facilitate optimizing public mental health intervention and control strategies. Objective This study aimed to build a social media-based pipeline that tracks public mental changes and use it to understand public mental health status regarding the pandemic. Methods This study used COVID-19-related tweets posted from February 2020 to April 2022. The tweets were downloaded using unique identifiers through the Twitter application programming interface. We created a lexicon of 4 mental health problems (depression, anxiety, insomnia, and addiction) to identify mental health-related tweets and developed a dictionary for identifying health care workers. We analyzed temporal and geographic distributions of public mental health status during the pandemic and further compared distributions among health care workers versus the general public, supplemented by topic modeling on their underlying foci. Finally, we used interrupted time series analysis to examine the statewide impact of a lockdown policy on public mental health in 12 states. Results We extracted 4,213,005 tweets related to mental health and COVID-19 from 2,316,817 users. Of these tweets, 2,161,357 (51.3%) were related to \"depression,\" whereas 1,923,635 (45.66%), 225,205 (5.35%), and 150,006 (3.56%) were related to \"anxiety,\" \"insomnia,\" and \"addiction,\" respectively. Compared to the general public, health care workers had higher risks of all 4 types of problems (all P Conclusions The impact of COVID-19 and the corresponding control measures on the public's mental status is dynamic and shows variability among different cohorts regarding disease types, occupations, and regional groups. Health agencies and policy makers should primarily focus on depression (reported by 51.3% of the tweets) and insomnia (which has had an ever-increasing trend since the beginning of the pandemic), especially among health care workers. Our pipeline timely tracks and analyzes public mental health changes, especially when primary studies and large-scale surveys are difficult to conduct.", "published_url": "https://doi.org/10.2196/39676", "crossref_type": "journal-article", "pmid": "36191167", "pmcid": "PMC9566822" }, { "title": "Low-resource Accent Classification in Geographically-Proximate Settings: A Forensic and Sociophonetics Perspective", "authors": [ "Qingcheng Zeng", "Dading Chong", "Peilin Zhou", "Jie Yang" ], "year": 2022, "venue": "Annual Conference of the International Speech Communication Association (INTERSPEECH)", "url": "https://arxiv.org/abs/2206.12759", "citation_page_url": "https://ylab.top/citation/low-resource-accent-classification-in-geographically-proximate-settings-a-forensic-and-sociophonetics-perspective/", "bibtex_url": "https://ylab.top/bib/low-resource-accent-classification-in-geographically-proximate-settings-a-forensic-and-sociophonetics-perspective.bib", "citation_key": "Zeng2022Lowresource", "bibtex_type": "inproceedings", "doi": "10.21437/interspeech.2022-11372", "metadata_sources": [ "Crossref", "arXiv" ], "metadata_match": { "crossref": 1.0 }, "preprint_url": "https://arxiv.org/abs/2206.12759", "pages": "5308-5312", "publisher": "ISCA", "abstract": "Accented speech recognition and accent classification are relatively under-explored research areas in speech technology. Recently, deep learning-based methods and Transformer-based pretrained models have achieved superb performances in both areas. However, most accent classification tasks focused on classifying different kinds of English accents and little attention was paid to geographically-proximate accent classification, especially under a low-resource setting where forensic speech science tasks usually encounter. In this paper, we explored three main accent modelling methods combined with two different classifiers based on 105 speaker recordings retrieved from five urban varieties in Northern England. Although speech representations generated from pretrained models generally have better performances in downstream classification, traditional methods like Mel Frequency Cepstral Coefficients (MFCCs) and formant measurements are equipped with specific strengths. These results suggest that in forensic phonetics scenario where data are relatively scarce, a simple modelling method and classifier could be competitive with state-of-the-art pretrained speech models as feature extractors, which could enhance a sooner estimation for the accent information in practices. Besides, our findings also cross-validated a new methodology in quantifying sociophonetic changes.", "published_url": "https://doi.org/10.21437/interspeech.2022-11372", "crossref_type": "proceedings-article", "arxiv_id": "2206.12759" }, { "title": "Natural Language Processing to Identify Abnormal Breast, Lung, and Cervical Cancer Screening Test Results from Unstructured Reports to Support Timely Follow-up", "authors": [ "Courtney J Diamond", "John Laurentiev", "Jie Yang", "Amy Wint", "Kimberly A Harris", "Tin H Dang", "Amrita Mecker", "Emily B Carpenter", "Anna N Tosteson", "Adam Wright" ], "year": 2022, "venue": "Studies in Health Technology and Informatics", "url": "https://ebooks.iospress.nl/doi/10.3233/SHTI220112", "citation_page_url": "https://ylab.top/citation/natural-language-processing-to-identify-abnormal-breast-lung-and-cervical-cancer-screening-test-results-from-unstructured-reports-to-support-timely-follow-up/", "bibtex_url": "https://ylab.top/bib/natural-language-processing-to-identify-abnormal-breast-lung-and-cervical-cancer-screening-test-results-from-unstructured-reports-to-support-timely-follow-up.bib", "citation_key": "Diamond2022Naturallanguage", "bibtex_type": "article", "doi": "10.3233/shti220112", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "pages": "433-437", "publisher": "IOS Press", "abstract": "Cancer screening and timely follow-up of abnormal results can reduce mortality. One barrier to follow-up is the failure to identify abnormal results. While EHRs have coded results for certain tests, cancer screening results are often stored in free-text reports, which limit capabilities for automated decision support. As part of the multilevel Follow-up of Cancer Screening (mFOCUS) trial, we developed and implemented a natural language processing (NLP) tool to assist with real-time detection of abnormal cancer screening test results (including mammograms, low-dose chest CT scans, and Pap smears) and identification of gynecological follow-up for higher risk abnormalities (i.e. colposcopy) from free-text reports. We demonstrate the integration and implementation of NLP, within the mFOCUS system, to improve the follow-up of abnormal cancer screening results in a large integrated healthcare system. The NLP pipelines have detected scenarios when guideline-recommended care was not delivered, in part because the provider mis-identified the text-based result reports.", "published_url": "https://doi.org/10.3233/shti220112", "crossref_type": "book-chapter", "pmid": "35673051", "pmcid": "PMC10792587" }, { "title": "Using Twitter Data to Understand Public Perceptions of Approved versus Off-label Use for COVID-19-related Medications", "authors": [ "Yining Hua", "Hang Jiang", "Shixu Lin", "Jie Yang", "Joseph M Plasek", "David W Bates", "Li Zhou" ], "year": 2022, "venue": "Journal of the American Medical Informatics Association", "url": "https://academic.oup.com/jamia/article/29/10/1668/6625661", "citation_page_url": "https://ylab.top/citation/using-twitter-data-to-understand-public-perceptions-of-approved-versus-off-label-use-for-covid-19-related-medications/", "bibtex_url": "https://ylab.top/bib/using-twitter-data-to-understand-public-perceptions-of-approved-versus-off-label-use-for-covid-19-related-medications.bib", "citation_key": "Hua2022Twitterdata", "bibtex_type": "article", "doi": "10.1093/jamia/ocac114", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "29", "issue": "10", "pages": "1668-1678", "publisher": "Oxford University Press (OUP)", "abstract": "Understanding public discourse on emergency use of unproven therapeutics is essential to monitor safe use and combat misinformation. We developed a natural language processing (NLP)-based pipeline to understand public perceptions of and stances on COVID-19-related drugs on Twitter across time. This retrospective study included 609,189 US-based tweets between January 29th, 2020 and November 30th, 2021 on four drugs that gained wide public attention during the COVID-19 pandemic: 1) Hydroxychloroquine and Ivermectin, drug therapies with anecdotal evidence; and 2) Molnupiravir and Remdesivir, FDA-approved treatment options for eligible patients. Time-trend analysis was used to understand the popularity and related events. Content and demographic analyses were conducted to explore potential rationales of people's stances on each drug. Time-trend analysis revealed that Hydroxychloroquine and Ivermectin received much more discussion than Molnupiravir and Remdesivir, particularly during COVID-19 surges. Hydroxychloroquine and Ivermectin were highly politicized, related to conspiracy theories, hearsay, celebrity effects, etc. The distribution of stance between the two major US political parties was significantly different (p<0.001); Republicans were much more likely to support Hydroxychloroquine (+55%) and Ivermectin (+30%) than Democrats. People with healthcare backgrounds tended to oppose Hydroxychloroquine (+7%) more than the general population; in contrast, the general population was more likely to support Ivermectin (+14%). We make all the data, code, and models available at https://github.com/ningkko/COVID-drug.", "published_url": "https://doi.org/10.1093/jamia/ocac114", "crossref_type": "journal-article" }, { "title": "Development and Validation of a Deep Learning Model for Earlier Detection of Cognitive Decline from Clinical Notes in Electronic Health Records", "authors": [ "Liqin Wang", "John Laurentiev", "Jie Yang", "Ying-Chih Lo", "Rebecca E. Amariglio", "Deborah Blacker", "Reisa A. Sperling", "Gad A. Marshall", "Li Zhou" ], "year": 2021, "venue": "JAMA Network Open", "url": "https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2786292", "citation_page_url": "https://ylab.top/citation/development-and-validation-of-a-deep-learning-model-for-earlier-detection-of-cognitive-decline-from-clinical-notes-in-electronic-health-records/", "bibtex_url": "https://ylab.top/bib/development-and-validation-of-a-deep-learning-model-for-earlier-detection-of-cognitive-decline-from-clinical-notes-in-electronic-health-records.bib", "citation_key": "Wang2021Developmentvalidation", "bibtex_type": "article", "doi": "10.1001/jamanetworkopen.2021.35174", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "4", "issue": "11", "pages": "e2135174", "publisher": "American Medical Association (AMA)", "abstract": "Importance Detecting cognitive decline earlier among older adults can facilitate enrollment in clinical trials and early interventions. Clinical notes in longitudinal electronic health records (EHRs) provide opportunities to detect cognitive decline earlier than it is noted in structured EHR fields as formal diagnoses. Objective To develop and validate a deep learning model to detect evidence of cognitive decline from clinical notes in the EHR. Design, setting, and participants Notes documented 4 years preceding the initial mild cognitive impairment (MCI) diagnosis were extracted from Mass General Brigham's Enterprise Data Warehouse for patients aged 50 years or older and with initial MCI diagnosis during 2019. The study was conducted from March 1, 2020, to June 30, 2021. Sections of notes for cognitive decline were labeled manually and 2 reference data sets were created. Data set I contained a random sample of 4950 note sections filtered by a list of keywords related to cognitive functions and was used for model training and testing. Data set II contained 2000 randomly selected sections without keyword filtering for assessing whether the model performance was dependent on specific keywords. Main outcomes and measures A deep learning model and 4 baseline models were developed and their performance was compared using the area under the receiver operating characteristic curve (AUROC) and area under the precision recall curve (AUPRC). Results Data set I represented 1969 patients (1046 [53.1%] women; mean [SD] age, 76.0 [13.3] years). Data set II comprised 1161 patients (619 [53.3%] women; mean [SD] age, 76.5 [10.2] years). With some overlap of patients deleted, the unique population was 2166. Cognitive decline was noted in 1453 sections (29.4%) in data set I and 69 sections (3.45%) in data set II. Compared with the 4 baseline models, the deep learning model achieved the best performance in both data sets, with AUROC of 0.971 (95% CI, 0.967-0.976) and AUPRC of 0.933 (95% CI, 0.921-0.944) for data set I and AUROC of 0.997 (95% CI, 0.994-0.999) and AUPRC of 0.929 (95% CI, 0.870-0.969) for data set II. Conclusions and relevance In this diagnostic study, a deep learning model accurately detected cognitive decline from clinical notes preceding MCI diagnosis and had better performance than keyword-based search and other machine learning models. These results suggest that a deep learning model could be used for earlier detection of cognitive decline in the EHRs.", "published_url": "https://doi.org/10.1001/jamanetworkopen.2021.35174", "crossref_type": "journal-article", "pmid": "34792589", "pmcid": "PMC8603078" }, { "title": "Assessing the Prognostic Significance of Tumor-Infiltrating Lymphocytes in Patients With Melanoma Using Pathologic Features Identified by Natural Language Processing", "authors": [ "Jie Yang", "John Lian", "Yen Po Chin", "Liqin Wang", "Anna Lian", "George F. Murphy", "Li Zhou" ], "year": 2021, "venue": "JAMA Network Open", "url": "https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2784429", "citation_page_url": "https://ylab.top/citation/assessing-the-prognostic-significance-of-tumor-infiltrating-lymphocytes-in-patients-with-melanoma-using-pathologic-features-identified-by-natural-language-processing/", "bibtex_url": "https://ylab.top/bib/assessing-the-prognostic-significance-of-tumor-infiltrating-lymphocytes-in-patients-with-melanoma-using-pathologic-features-identified-by-natural-language-processing.bib", "citation_key": "Yang2021Assessingprognostic", "bibtex_type": "article", "doi": "10.1001/jamanetworkopen.2021.26337", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "4", "issue": "9", "pages": "e2126337", "publisher": "American Medical Association (AMA)", "abstract": "Importance Although tumor-infiltrating lymphocytes (TILs) are an important histopathologic characteristic reflecting host immune response in patients with melanoma, their prognostic value remains controversial. Because manual review of medical records is labor intensive, a survival analysis using a large patient cohort with comprehensive clinical and histopathologic characteristics is lacking. Objective To assess the prognostic significance of TILs among patients with cutaneous melanoma using a large cohort established through natural language processing (NLP) algorithms. Design, setting, and participants This retrospective cohort study analyzed the medical records of 14 436 patients with cutaneous melanoma at Brigham and Women's Hospital between June 1, 2004, and December 31, 2019. Patients were followed up to death or censored at their last clinical visit. Main outcome and measures The primary outcome was overall survival (OS). Survival analysis was conducted using Kaplan-Meier curves, the log-rank test, and Cox proportional hazards regression analysis. Results A total of 14 436 patients with cutaneous melanoma were identified in the institution's pathology information system. Using NLP, we established a study cohort of 2624 patients (1462 men [55.7%]; median age, 61 years [interquartile range, 50-72 years]) who had vertical growth phase melanoma with TIL status scored. Absent TILs were identified in 434 patients (16.5%), nonbrisk TILs in 1916 patients (73.0%), and brisk TILs in 274 patients (10.4%). The 5-year survival rate was 71.0% (95% CI, 65.5%-76.9%) among patients with an absence of TILs, 73.8% (95% CI, 71.1%-76.5%) among patients with nonbrisk TILs, and 85.2% (95% CI, 80.0%-90.7%) among patients with brisk TILs. Brisk TILs were significantly associated with improved OS (adjusted hazard ratio, 0.63; 95% CI, 0.42-0.95; P = .03; 14.2% OS advantage at 5 years), and nonbrisk TILs were not associated with improved OS (adjusted hazard ratio, 0.87; 95% CI, 0.68-1.11; P = .25), compared with the absence of TILs. Conclusions and relevance This study provides evidence based on a large patient cohort from a single institution that suggests that brisk TILs represent an independent prognostic factor for OS among patients with primary cutaneous melanoma. The study also suggests that NLP is a highly efficient tool to facilitate large-scale analyses that involve free-text clinical data.", "published_url": "https://doi.org/10.1001/jamanetworkopen.2021.26337", "crossref_type": "journal-article", "pmid": "34550383", "pmcid": "PMC8459191" }, { "title": "Sentiment Correlation in Financial News Networks and Associated Market Movements", "authors": [ "Xingchen Wan", "Jie Yang", "Slavi Marinov", "Jan-Peter Calliess", "Stefan Zohren", "Xiaowen Dong" ], "year": 2021, "venue": "Scientific Reports", "url": "https://www.nature.com/articles/s41598-021-82338-3", "citation_page_url": "https://ylab.top/citation/sentiment-correlation-in-financial-news-networks-and-associated-market-movements/", "bibtex_url": "https://ylab.top/bib/sentiment-correlation-in-financial-news-networks-and-associated-market-movements.bib", "citation_key": "Wan2021Sentimentcorrelation", "bibtex_type": "article", "doi": "10.1038/s41598-021-82338-6", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "11", "issue": "1", "pages": "3062", "publisher": "Springer Science and Business Media LLC", "abstract": "In an increasingly connected global market, news sentiment towards one company may not only indicate its own market performance, but can also be associated with a broader movement on the sentiment and performance of other companies from the same or even different sectors. In this paper, we apply NLP techniques to understand news sentiment of 87 companies among the most reported on Reuters for a period of 7 years. We investigate the propagation of such sentiment in company networks and evaluate the associated market movements in terms of stock price and volatility. Our results suggest that, in certain sectors, strong media sentiment towards one company may indicate a significant change in media sentiment towards related companies measured as neighbours in a financial network constructed from news co-occurrence. Furthermore, there exists a weak but statistically significant association between strong media sentiment and abnormal market return as well as volatility. Such an association is more significant at the level of individual companies, but nevertheless remains visible at the level of sectors or groups of companies.", "published_url": "https://doi.org/10.1038/s41598-021-82338-6", "crossref_type": "journal-article", "pmid": "33542292", "pmcid": "PMC7862280" }, { "title": "Multilevel Follow-up of Cancer Screening (mFOCUS): Protocol for a Multilevel Intervention to Improve the Follow-up of Abnormal Cancer Screening Test Results", "authors": [ "Jennifer S Haas", "Steven J Atlas", "Adam Wright", "E John Orav", "David G Aman", "Erica S Breslau", "Timothy E Burdick", "Emily Carpenter", "Frank Chang", "Tin Dang", "Jie Yang", "Li Zhou", "Anna Tosteson" ], "year": 2021, "venue": "Contemporary Clinical Trials", "url": "https://www.sciencedirect.com/science/article/pii/S155171442100269X", "citation_page_url": "https://ylab.top/citation/multilevel-follow-up-of-cancer-screening-mfocus-protocol-for-a-multilevel-intervention-to-improve-the-follow-up-of-abnormal-cancer-screening-test-results/", "bibtex_url": "https://ylab.top/bib/multilevel-follow-up-of-cancer-screening-mfocus-protocol-for-a-multilevel-intervention-to-improve-the-follow-up-of-abnormal-cancer-screening-test-results.bib", "citation_key": "Haas2021Multilevelfollow", "bibtex_type": "article", "doi": "10.1016/j.cct.2021.106533", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "109", "pages": "106533", "publisher": "Elsevier BV", "abstract": "Introduction While substantial attention is focused on the delivery of routine preventive cancer screening, less attention has been paid to systematically ensuring that there is timely follow-up of abnormal screening test results. Barriers to completion of timely follow-up occur at the patient, provider, care team and system levels. Methods In this pragmatic cluster randomized controlled trial, primary care sites in three networks are randomized to one of four arms: (1) standard care, (2) \"visit-based\" reminders that appear in a patient's electronic health record (EHR) when it is accessed by either patient or providers (3) visit based reminders with population health outreach, and (4) visit based reminders, population health outreach, and patient navigation with systematic screening and referral to address social barriers to care. Eligible patients in participating practices are those overdue for follow-up of an abnormal results on breast, cervical, colorectal and lung cancer screening tests. Results The primary outcome is whether an individual receives follow-up, specific to the organ type and screening abnormality, within 120 days of becoming eligible for the trial. Secondary outcomes assess the effect of intervention components on the patient and provider experience of obtaining follow-up care and the delivery of the intervention components. Conclusions This trial will provide evidence for the role of a multilevel intervention on improving the follow-up of abnormal cancer screening test results. We will also specifically assess the relative impact of the components of the intervention, compared to standard care. Trial registration ClinicalTrials.gov NCT03979495.", "published_url": "https://doi.org/10.1016/j.cct.2021.106533", "crossref_type": "journal-article", "pmid": "34375748", "pmcid": "PMC8900526" }, { "title": "Development and Validation of a Deep Learning Model for Detection of Allergic Reactions Using Safety Event Reports Across Hospitals", "authors": [ "Jie Yang", "Liqin Wang", "Neelam A. Phadke", "Paige G. Wickner", "Christian M. Mancini", "Kimberly G. Blumenthal", "Li Zhou" ], "year": 2020, "venue": "JAMA Network Open", "url": "https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2772992", "citation_page_url": "https://ylab.top/citation/development-and-validation-of-a-deep-learning-model-for-detection-of-allergic-reactions-using-safety-event-reports-across-hospitals/", "bibtex_url": "https://ylab.top/bib/development-and-validation-of-a-deep-learning-model-for-detection-of-allergic-reactions-using-safety-event-reports-across-hospitals.bib", "citation_key": "Yang2020Developmentvalidation", "bibtex_type": "article", "doi": "10.1001/jamanetworkopen.2020.22836", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "pmid": "33196805", "pmcid": "PMC7670315", "volume": "3", "issue": "11", "pages": "e2022836", "publisher": "American Medical Association (AMA)", "abstract": "Importance Although critical to patient safety, health care-related allergic reactions are challenging to identify and monitor. Objective To develop a deep learning model to identify allergic reactions in the free-text narrative of hospital safety reports and evaluate its generalizability, efficiency, productivity, and interpretability. Design, setting, and participants This cross-sectional study analyzed hospital safety reports filed between May 2004 and January 2019 at Brigham and Women's Hospital and between April 2006 and June 2018 at Massachusetts General Hospital in Boston. Training and validating a deep learning model involved extracting safety reports using 101 expert-curated keywords from Massachusetts General Hospital (data set I). The model was then evaluated on 3 data sets: reports without keywords (data set II), reports from a different time frame (data set III), and reports from a different hospital (Brigham and Women's Hospital; data set IV). Statistical analyses were performed between March 1, 2019, and July 18, 2020. Main outcomes and measures The area under the receiver operating characteristic curve and area under the precision-recall curve were used on data set I. The precision at top-k was used on data sets II to IV. Results A total of 299 028 safety reports with 172 854 patients were included. Of these patients, 86 544 were women (50.1%) and the median (interquartile range [IQR]) age was 59.7 (43.8-71.6) years. The deep learning model achieved an area under the receiver operating characteristic curve of 0.979 (95% CI, 0.973-0.985) and an area under the precision-recall curve of 0.809 (95% CI, 0.773-0.845). The model achieved precisions at the top 100 model-identified cases of 0.930 in data set II, 0.960 in data set III, and 0.990 in data set IV. Compared with the keyword-search approach, the deep learning model reduced the number of cases for manual review by 63.8% and identified 24.2% more cases of confirmed allergic reactions. The model highlighted important words (eg, rash, hives, and Benadryl) in prediction and extended the list of expert-curated keywords through an attention layer. Conclusions and relevance This study showed that a deep learning model can accurately and efficiently identify allergic reactions using free-text narratives written by a variety of health care professionals. This model could be used to improve allergy care, potentially enabling real-time event surveillance and guidance for medical errors and system improvement.", "published_url": "https://doi.org/10.1001/jamanetworkopen.2020.22836", "crossref_type": "journal-article" }, { "title": "Deep Learning to Detect Allergy Events from Hospital Safety Reports", "authors": [ "Jie Yang", "Liqin Wang", "Neelam A Phadke", "Paige G Wickner", "Christian M Mancini", "Kimberly G Blumenthal", "Li Zhou" ], "year": 2020, "venue": "American Medical Informatics Association Annual Symposium (AMIA)", "url": null, "citation_page_url": "https://ylab.top/citation/deep-learning-to-detect-allergy-events-from-hospital-safety-reports/", "bibtex_url": "https://ylab.top/bib/deep-learning-to-detect-allergy-events-from-hospital-safety-reports.bib", "citation_key": "Yang2020Deeplearning", "bibtex_type": "inproceedings", "doi": null }, { "title": "Implementing an IT-Based Intervention to Improve Follow-up Rates of Abnormal Cancer Screening Results: the mFOCUS Trial", "authors": [ "Courtney Diamond", "Steven Atlas", "Tin Dang", "Jie Yang", "Li Zhou", "Sanja Percac-Lima", "Amy Wint", "Kimberly Harris", "John Orav", "Erica Breslau", "Shoshana Hort", "Anna Tosteson", "Jennifer Haas", "Adam Wright" ], "year": 2020, "venue": "American Medical Informatics Association Annual Symposium (AMIA)", "url": null, "citation_page_url": "https://ylab.top/citation/implementing-an-it-based-intervention-to-improve-follow-up-rates-of-abnormal-cancer-screening-results-the-mfocus-trial/", "bibtex_url": "https://ylab.top/bib/implementing-an-it-based-intervention-to-improve-follow-up-rates-of-abnormal-cancer-screening-results-the-mfocus-trial.bib", "citation_key": "Diamond2020Implementingit", "bibtex_type": "inproceedings", "doi": null }, { "title": "Allergic Reactions in Two Academic Medical Centers", "authors": [ "Neelam A. Phadke", "Li Zhou", "Christian M. Mancini", "Jie Yang", "Liqin Wang", "Paige Wickner", "Xiaoqing Fu", "Kimberly G. Blumenthal" ], "year": 2020, "venue": "Journal of General Internal Medicine", "url": "https://link.springer.com/article/10.1007/s11606-020-06190-6", "citation_page_url": "https://ylab.top/citation/allergic-reactions-in-two-academic-medical-centers/", "bibtex_url": "https://ylab.top/bib/allergic-reactions-in-two-academic-medical-centers.bib", "citation_key": "Phadke2020Allergicreactions", "bibtex_type": "article", "doi": "10.1007/s11606-020-06190-6", "metadata_sources": [ "Crossref", "Europe PMC" ], "metadata_match": { "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "36", "issue": "6", "pages": "1814-1817", "publisher": "Springer Science and Business Media LLC", "published_url": "https://doi.org/10.1007/s11606-020-06190-6", "published_year": 2021, "crossref_type": "journal-article", "pmid": "32959347", "pmcid": "PMC8175601" }, { "title": "Disparities in Outpatient Visits for Mental Health and/or Substance Use Disorders During the COVID-19 Surge and Partial Reopening in Massachusetts", "authors": [ "Jie Yang", "Mary B. Landrum", "Li Zhou", "Alisa B Busch" ], "year": 2020, "venue": "General Hospital Psychiatry", "url": "https://www.sciencedirect.com/science/article/pii/S0163834320301407", "citation_page_url": "https://ylab.top/citation/disparities-in-outpatient-visits-for-mental-health-and-or-substance-use-disorders-during-the-covid-19-surge-and-partial-reopening-in-massachusetts/", "bibtex_url": "https://ylab.top/bib/disparities-in-outpatient-visits-for-mental-health-and-or-substance-use-disorders-during-the-covid-19-surge-and-partial-reopening-in-massachusetts.bib", "citation_key": "Yang2020Disparitiesoutpatient", "bibtex_type": "article", "doi": "10.1016/j.genhosppsych.2020.09.004", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 0.9896, "crossref": 0.9896 }, "pmid": "33091782", "pmcid": "PMC7550185", "volume": "67", "pages": "100-106", "publisher": "Elsevier BV", "published_url": "https://doi.org/10.1016/j.genhosppsych.2020.09.004", "published_title": "Disparities in outpatient visits for mental health and/or substance use disorders during the COVID surge and partial reopening in Massachusetts", "crossref_type": "journal-article" }, { "title": "Lattice LSTM for Chinese Sentence Representation", "authors": [ "Yue Zhang", "Yile Wang", "Jie Yang" ], "year": 2020, "venue": "IEEE/ACM Transactions on Audio, Speech, and Language Processing", "url": "https://ieeexplore.ieee.org/document/9082853", "citation_page_url": "https://ylab.top/citation/lattice-lstm-for-chinese-sentence-representation/", "bibtex_url": "https://ylab.top/bib/lattice-lstm-for-chinese-sentence-representation.bib", "citation_key": "Zhang2020Latticelstm", "bibtex_type": "article", "doi": "10.1109/taslp.2020.2991544", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "volume": "28", "pages": "1506-1519", "publisher": "Institute of Electrical and Electronics Engineers (IEEE)", "published_url": "https://doi.org/10.1109/taslp.2020.2991544", "crossref_type": "journal-article" }, { "title": "Subword Encoding in Lattice LSTM for Chinese Word Segmentation", "authors": [ "Jie Yang", "Yue Zhang", "Shuailong Liang" ], "year": 2019, "venue": "Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL)", "url": "https://aclanthology.org/N19-1278/", "citation_page_url": "https://ylab.top/citation/subword-encoding-in-lattice-lstm-for-chinese-word-segmentation/", "bibtex_url": "https://ylab.top/bib/subword-encoding-in-lattice-lstm-for-chinese-word-segmentation.bib", "citation_key": "Yang2019Subwordencoding", "bibtex_type": "inproceedings", "doi": "10.18653/v1/n19-1278", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "pages": "2720-2725", "publisher": "Association for Computational Linguistics", "published_url": "https://doi.org/10.18653/v1/n19-1278", "crossref_type": "proceedings-article" }, { "title": "Mining Clinical Phrases from Nursing Notes to Discover Risk Factors of Patient Deterioration", "authors": [ "Zfania T. Korach", "Jie Yang", "Sarah Collins Rossetti", "Kenrick D. Cato", "Min-Jeoung Kang", "Christopher Knaplund", "Kumiko O. Schnock", "Jose P. Garcia", "Haomiao Jia", "Jessica M. Schwartz", "Li Zhou" ], "year": 2019, "venue": "International Journal of Medical Informatics", "url": "https://www.sciencedirect.com/science/article/pii/S1386505619309682", "citation_page_url": "https://ylab.top/citation/mining-clinical-phrases-from-nursing-notes-to-discover-risk-factors-of-patient-deterioration/", "bibtex_url": "https://ylab.top/bib/mining-clinical-phrases-from-nursing-notes-to-discover-risk-factors-of-patient-deterioration.bib", "citation_key": "Korach2019Miningclinical", "bibtex_type": "article", "doi": "10.1016/j.ijmedinf.2019.104053", "metadata_sources": [ "ORCID", "Crossref", "Europe PMC" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0, "europe_pmc": 1.0 }, "volume": "135", "pages": "104053", "publisher": "Elsevier BV", "abstract": "Objective Early identification and treatment of patient deterioration is crucial to improving clinical outcomes. To act, hospital rapid response (RR) teams often rely on nurses' clinical judgement typically documented narratively in the electronic health record (EHR). We developed a data-driven, unsupervised method to discover potential risk factors of RR events from nursing notes. Methods We applied multiple natural language processing methods, including language modelling, word embeddings, and two phrase mining methods (TextRank and NC-Value), to identify quality phrases that represent clinical entities from unannotated nursing notes. TextRank was used to determine the important word-sequences in each note. NC-Value was then used to globally rank the locally-important sequences across the whole corpus. We evaluated our method both on its accuracy compared to human judgement and on the ability of the mined phrases to predict a clinical outcome, RR event hazard. Results When applied to 61,740 hospital encounters with 1,067 RR events and 778,955 notes, our method achieved an average precision of 0.590 to 0.764 (when excluding numeric tokens). Time-dependent covariates Cox model using the phrases achieved a concordance index of 0.739. Clustering the phrases revealed clinical concepts significantly associated with RR event hazard. Discussion Our findings demonstrate that our minimal-annotation, unsurprised method can rapidly mine quality phrases from a large amount of nursing notes, and these identified phrases are useful for downstream tasks, such as clinical outcome predication and risk factor identification.", "published_url": "https://doi.org/10.1016/j.ijmedinf.2019.104053", "published_year": 2020, "crossref_type": "journal-article", "pmid": "31884312", "pmcid": "PMC7103062" }, { "title": "From Genesis to Creole Language: Transfer Learning for Singlish Universal Dependencies Parsing and POS Tagging", "authors": [ "Hongming Wang", "Jie Yang", "Yue Zhang" ], "year": 2019, "venue": "ACM Transactions on Asian and Low-Resource Language Information Processing", "url": "https://dl.acm.org/doi/10.1145/3321128", "citation_page_url": "https://ylab.top/citation/from-genesis-to-creole-language-transfer-learning-for-singlish-universal-dependencies-parsing-and-pos-tagging/", "bibtex_url": "https://ylab.top/bib/from-genesis-to-creole-language-transfer-learning-for-singlish-universal-dependencies-parsing-and-pos-tagging.bib", "citation_key": "Wang2019Genesiscreole", "bibtex_type": "article", "doi": "10.1145/3321128", "metadata_sources": [ "ORCID" ], "metadata_match": { "orcid": 1.0 } }, { "title": "Design Challenges and Misconceptions in Neural Sequence Labeling", "authors": [ "Jie Yang", "Shuailong Liang", "Yue Zhang" ], "year": 2018, "venue": "International Conference on Computational Linguistics (COLING)", "url": "https://aclanthology.org/C18-1327/", "citation_page_url": "https://ylab.top/citation/design-challenges-and-misconceptions-in-neural-sequence-labeling/", "bibtex_url": "https://ylab.top/bib/design-challenges-and-misconceptions-in-neural-sequence-labeling.bib", "citation_key": "Yang2018Designchallenges", "bibtex_type": "inproceedings", "doi": "10.18653/v1/C18-1327", "metadata_sources": [ "ORCID" ], "metadata_match": { "orcid": 1.0 } }, { "title": "NCRF++: An Open-source Neural Sequence Labeling Toolkit", "authors": [ "Jie Yang", "Yue Zhang" ], "year": 2018, "venue": "Annual Meeting of the Association for Computational Linguistics (ACL), Demo", "url": "https://aclanthology.org/P18-4013/", "citation_page_url": "https://ylab.top/citation/ncrf-an-open-source-neural-sequence-labeling-toolkit/", "bibtex_url": "https://ylab.top/bib/ncrf-an-open-source-neural-sequence-labeling-toolkit.bib", "citation_key": "Yang2018Ncrfopen", "bibtex_type": "inproceedings", "doi": "10.18653/v1/p18-4013", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "code_url": "https://github.com/jiesutd/NCRFpp", "pages": "74-79", "publisher": "Association for Computational Linguistics", "published_url": "https://doi.org/10.18653/v1/p18-4013", "crossref_type": "proceedings-article" }, { "title": "Chinese NER Using Lattice LSTM", "authors": [ "Yue Zhang", "Jie Yang" ], "year": 2018, "venue": "Annual Meeting of the Association for Computational Linguistics (ACL)", "url": "https://aclanthology.org/P18-1144/", "citation_page_url": "https://ylab.top/citation/chinese-ner-using-lattice-lstm/", "bibtex_url": "https://ylab.top/bib/chinese-ner-using-lattice-lstm.bib", "citation_key": "Zhang2018Chinesener", "bibtex_type": "inproceedings", "doi": "10.18653/v1/p18-1144", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "code_url": "https://github.com/jiesutd/LatticeLSTM", "pages": "1554-1564", "publisher": "Association for Computational Linguistics", "published_url": "https://doi.org/10.18653/v1/p18-1144", "crossref_type": "proceedings-article" }, { "title": "YEDDA: A Lightweight Collaborative Text Span Annotation Tool", "authors": [ "Jie Yang", "Yue Zhang", "Linwei Li", "Xingxuan Li" ], "year": 2018, "venue": "Annual Meeting of the Association for Computational Linguistics (ACL), Demo", "url": "https://aclanthology.org/P18-4006/", "citation_page_url": "https://ylab.top/citation/yedda-a-lightweight-collaborative-text-span-annotation-tool/", "bibtex_url": "https://ylab.top/bib/yedda-a-lightweight-collaborative-text-span-annotation-tool.bib", "citation_key": "Yang2018Yeddalightweight", "bibtex_type": "inproceedings", "doi": "10.18653/v1/p18-4006", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "code_url": "https://github.com/jiesutd/YEDDA", "pages": "31-36", "publisher": "Association for Computational Linguistics", "published_url": "https://doi.org/10.18653/v1/p18-4006", "crossref_type": "proceedings-article" }, { "title": "Neural Word Segmentation with Rich Pretraining", "authors": [ "Jie Yang", "Yue Zhang", "Fei Dong" ], "year": 2017, "venue": "Annual Meeting of the Association for Computational Linguistics (ACL)", "url": "https://aclanthology.org/P17-1078/", "citation_page_url": "https://ylab.top/citation/neural-word-segmentation-with-rich-pretraining/", "bibtex_url": "https://ylab.top/bib/neural-word-segmentation-with-rich-pretraining.bib", "citation_key": "Yang2017Neuralword", "bibtex_type": "inproceedings", "doi": "10.18653/v1/p17-1078", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "pages": "839-849", "publisher": "Association for Computational Linguistics", "published_url": "https://doi.org/10.18653/v1/p17-1078", "crossref_type": "proceedings-article" }, { "title": "Universal Dependencies Parsing for Colloquial Singaporean English", "authors": [ "Hongmin Wang", "Yue Zhang", "Guang Yow Leonard Chan", "Jie Yang", "Hai Leong Chieu" ], "year": 2017, "venue": "Annual Meeting of the Association for Computational Linguistics (ACL)", "url": "https://aclanthology.org/P17-1159/", "citation_page_url": "https://ylab.top/citation/universal-dependencies-parsing-for-colloquial-singaporean-english/", "bibtex_url": "https://ylab.top/bib/universal-dependencies-parsing-for-colloquial-singaporean-english.bib", "citation_key": "Wang2017Universaldependencies", "bibtex_type": "inproceedings", "doi": "10.18653/v1/p17-1159", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "pages": "1732-1744", "publisher": "Association for Computational Linguistics", "published_url": "https://doi.org/10.18653/v1/p17-1159", "crossref_type": "proceedings-article" }, { "title": "Attention-based Recurrent Convolutional Neural Network for Automatic Essay Scoring", "authors": [ "Fei Dong", "Yue Zhang", "Jie Yang" ], "year": 2017, "venue": "Conference on Computational Natural Language Learning (CoNLL)", "url": "https://aclanthology.org/K17-1017/", "citation_page_url": "https://ylab.top/citation/attention-based-recurrent-convolutional-neural-network-for-automatic-essay-scoring/", "bibtex_url": "https://ylab.top/bib/attention-based-recurrent-convolutional-neural-network-for-automatic-essay-scoring.bib", "citation_key": "Dong2017Attentionbased", "bibtex_type": "inproceedings", "doi": "10.18653/v1/k17-1017", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "pages": "153-162", "publisher": "Association for Computational Linguistics", "published_url": "https://doi.org/10.18653/v1/k17-1017", "crossref_type": "proceedings-article" }, { "title": "Neural Reranking for Named Entity Recognition", "authors": [ "Jie Yang", "Yue Zhang", "Fei Dong" ], "year": 2017, "venue": "Recent Advances in Natural Language Processing (RANLP)", "url": "https://aclanthology.org/R17-1101/", "citation_page_url": "https://ylab.top/citation/neural-reranking-for-named-entity-recognition/", "bibtex_url": "https://ylab.top/bib/neural-reranking-for-named-entity-recognition.bib", "citation_key": "Yang2017Neuralreranking", "bibtex_type": "inproceedings", "doi": "10.26615/978-954-452-049-6_101", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "pages": "784-792", "publisher": "Incoma Ltd. Shoumen, Bulgaria", "published_url": "https://doi.org/10.26615/978-954-452-049-6_101", "crossref_type": "proceedings-article" }, { "title": "Combining Discrete and Neural Features for Sequence Labeling", "authors": [ "Jie Yang", "Zhiyang Teng", "Meishan Zhang", "Yue Zhang" ], "year": 2016, "venue": "International Conference on Intelligent Text Processing and Computational Linguistics (CICLing)", "url": "https://arxiv.org/abs/1708.07279", "citation_page_url": "https://ylab.top/citation/combining-discrete-and-neural-features-for-sequence-labeling/", "bibtex_url": "https://ylab.top/bib/combining-discrete-and-neural-features-for-sequence-labeling.bib", "citation_key": "Yang2016Combiningdiscrete", "bibtex_type": "inproceedings", "doi": "10.1007/978-3-319-75477-2_9", "metadata_sources": [ "ORCID", "Crossref", "arXiv" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "preprint_url": "https://arxiv.org/abs/1708.07279", "pages": "140-154", "publisher": "Springer International Publishing", "abstract": "Neural network models have recently received heated research attention in the natural language processing community. Compared with traditional models with discrete features, neural models have two main advantages. First, they take low-dimensional, real-valued embedding vectors as inputs, which can be trained over large raw data, thereby addressing the issue of feature sparsity in discrete models. Second, deep neural networks can be used to automatically combine input features, and including non-local features that capture semantic patterns that cannot be expressed using discrete indicator features. As a result, neural network models have achieved competitive accuracies compared with the best discrete models for a range of NLP tasks. On the other hand, manual feature templates have been carefully investigated for most NLP tasks over decades and typically cover the most useful indicator pattern for solving the problems. Such information can be complementary the features automatically induced from neural networks, and therefore combining discrete and neural features can potentially lead to better accuracy compared with models that leverage discrete or neural features only. In this paper, we systematically investigate the effect of discrete and neural feature combination for a range of fundamental NLP tasks based on sequence labeling, including word segmentation, POS tagging and named entity recognition for Chinese and English, respectively. Our results on standard benchmarks show that state-of-the-art neural models can give accuracies comparable to the best discrete models in the literature for most tasks and combing discrete and neural features unanimously yield better results.", "published_url": "https://doi.org/10.1007/978-3-319-75477-2_9", "published_year": 2018, "crossref_type": "book-chapter", "arxiv_id": "1708.07279" }, { "title": "LibN3L: A Lightweight Package for Neural NLP", "authors": [ "Meishan Zhang", "Jie Yang", "Zhiyang Teng", "Yue Zhang" ], "year": 2016, "venue": "Language Resources and Evaluation Conference (LREC)", "url": "https://aclanthology.org/L16-1034/", "citation_page_url": "https://ylab.top/citation/libn3l-a-lightweight-package-for-neural-nlp/", "bibtex_url": "https://ylab.top/bib/libn3l-a-lightweight-package-for-neural-nlp.bib", "citation_key": "Zhang2016Libn3llightweight", "bibtex_type": "inproceedings", "doi": "10.63317/3nqgpgnhe8bz", "metadata_sources": [ "ORCID", "Crossref" ], "metadata_match": { "orcid": 1.0, "crossref": 1.0 }, "pages": "225-229", "publisher": "European Language Resources Association (ELRA)", "published_url": "https://doi.org/10.63317/3nqgpgnhe8bz", "crossref_type": "proceedings-article" }, { "title": "Effects of Defects and Thermal Treatment on the Properties of Graphene", "authors": [ "Kunpeng Jia", "Yajuan Su", "Yang Chen", "Jun Luo", "Jie Yang", "Peng Lv", "Zihan Zhang", "Huilong Zhu", "Chao Zhao", "Tianchun Ye" ], "year": 2015, "venue": "Vacuum", "url": "https://www.sciencedirect.com/science/article/pii/S0042207X15000949", "citation_page_url": "https://ylab.top/citation/effects-of-defects-and-thermal-treatment-on-the-properties-of-graphene/", "bibtex_url": "https://ylab.top/bib/effects-of-defects-and-thermal-treatment-on-the-properties-of-graphene.bib", "citation_key": "Jia2015Effectsdefects", "bibtex_type": "article", "doi": "10.1016/j.vacuum.2015.03.003", "metadata_sources": [ "Crossref" ], "metadata_match": { "crossref": 1.0 }, "volume": "116", "pages": "90-95", "publisher": "Elsevier BV", "published_url": "https://doi.org/10.1016/j.vacuum.2015.03.003", "crossref_type": "journal-article" }, { "title": "Hysteresis Analysis of Graphene Transistor Under Repeated Test and Gate Voltage Stress", "authors": [ "Jie Yang", "Kunpeng Jia", "Yajuan Su", "Yang Chen", "Chao Zhao" ], "year": 2014, "venue": "Journal of Semiconductors", "url": "https://iopscience.iop.org/article/10.1088/1674-4926/35/9/094003", "citation_page_url": "https://ylab.top/citation/hysteresis-analysis-of-graphene-transistor-under-repeated-test-and-gate-voltage-stress/", "bibtex_url": "https://ylab.top/bib/hysteresis-analysis-of-graphene-transistor-under-repeated-test-and-gate-voltage-stress.bib", "citation_key": "Yang2014Hysteresisanalysis", "bibtex_type": "article", "doi": "10.1088/1674-4926/35/9/094003", "metadata_sources": [ "Crossref" ], "metadata_match": { "crossref": 1.0 }, "volume": "35", "issue": "9", "pages": "094003", "publisher": "IOP Publishing", "published_url": "https://doi.org/10.1088/1674-4926/35/9/094003", "crossref_type": "journal-article" }, { "title": "Stability Analysis of a Back-gate Graphene Transistor in Air Environment", "authors": [ "Kunpeng Jia", "Jie Yang", "Yajuan Su", "Pengfei Nie", "Jian Zhong", "Qingqing Liang", "Huilong Zhu" ], "year": 2013, "venue": "Journal of Semiconductors", "url": "https://iopscience.iop.org/article/10.1088/1674-4926/34/8/084004", "citation_page_url": "https://ylab.top/citation/stability-analysis-of-a-back-gate-graphene-transistor-in-air-environment/", "bibtex_url": "https://ylab.top/bib/stability-analysis-of-a-back-gate-graphene-transistor-in-air-environment.bib", "citation_key": "Jia2013Stabilityanalysis", "bibtex_type": "article", "doi": "10.1088/1674-4926/34/8/084004", "metadata_sources": [ "Crossref" ], "metadata_match": { "crossref": 1.0 }, "volume": "34", "issue": "8", "pages": "084004", "publisher": "IOP Publishing", "published_url": "https://doi.org/10.1088/1674-4926/34/8/084004", "crossref_type": "journal-article" } ] }