Development and validation of venous thromboembolism–bidirectional encoder representations from transformers (VTE-BERT) natural language processing model
Journal of Thrombosis and Haemostasis · 2026-07 · 24 (7) · 2522-2531
DOI: 10.1016/j.jtha.2025.07.021 ↗Kurz gefasst
Validierungsstudie in der Zeitschrift Journal of Thrombosis and Haemostasis, 2026-07. Verfasst von Omid Jafari und 14 weiteren Autor:innen. Beteiligte Einrichtung: Section of Hematology-Oncology, Department of Medicine, Baylor College of Medicine, Houston, Texas, USA. Hauptthemen laut MEDLINE: Natural Language Processing, Venous Thromboembolism, Electronic Health Records, Data Mining. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: NHLBI NIH HHS, NIH HHS, Department of Veterans Affairs.
Themen: Thromboseprophylaxe
Abstract (Originalsprache)
BACKGROUND: Accurate and rapid phenotyping of venous thromboembolism (VTE) in longitudinal studies is important. A natural language processing (NLP) tool externally validated in representative patients is lacking. OBJECTIVES: To train and validate an efficient NLP model to detect incident VTE event. METHODS: We designed a novel NLP platform, NLPMed, to assist thrombosis researchers with data preprocessing, phenotype annotation, language model finetuning, and NLP application. Using clinical notes, discharge summaries, and radiology reports from patients with cancer at 2 healthcare institutions, we finetuned Bio_Clinical Bidirectional Encoder Representations from Transformers (BERT) to develop VTE-BERT. The new model was trained to detect acute VTE events and their anatomical locations longitudinally. We internally and externally validated the model's performance in 2 randomly sampled cohorts of patients with advanced cancer. RESULTS: The training cohort consisted of 715 patients and 14 013 annotated notes with ≥1 VTE keyword from the Harris Health System. The internal validation cohort included 400 additional patients with 7190 VTE keyword-containing notes from Harris Health System. The external validation cohort included 400 patients with 7371 VTE keyword-containing notes from the national Veterans Affairs healthcare system. VTE-BERT was trained until it reached a precision of 95% and recall of 98% on the patient level. Using independent datasets, the model achieved precision and recall of 95% and 91% in internal validation and of 85% and 92% in external validation. CONCLUSION: We trained and externally validated an efficient NLP model to detect incident VTE events longitudinally. We believe its adoption will accelerate thrombosis research by improving VTE detection at scale and decreasing the time and expense involved with manual chart review in big data epidemiological studies.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Section of Hematology-Oncology, Department of Medicine, Baylor College of Medicine, Houston, Texas, USA.
- Division of Hematology and Oncology, Fred Hutch Cancer Center, University of Washington, Seattle, Washington, USA.
- McGovern Medical School, University of Texas Health Science Center at Houston, Houston, Texas, USA.
- Section of Hematology, Department of Medicine, Yale School of Medicine, New Haven, Connecticut, USA.
- Tegria Services Group, Dallas, Texas, USA; Harris Health System, Harris County, Texas, USA.
- Institute for Clinical and Translational Research, Baylor College of Medicine, Houston, Texas, USA.
Förderung
- NHLBI NIH HHS
- NIH HHS
- Department of Veterans Affairs
- American Society of Hematology
- National Heart, Lung, and Blood Institute
- Cancer Prevention and Research Institute of Texas
- Conquer Cancer Foundation
- American Society of Clinical Oncology
Zitieren
APA
Jafari, O., Ma, S., Lam, B. D., Jiang, J. Y., Zhou, E., Ranjan, M., Ryu, J., Bandyo, R., Maghsoudi, A., Peng, B., Amos, C. I., Oluyomi, A., Fillmore, N. R., La, J., & Li, A. (2026). Development and validation of venous thromboembolism–bidirectional encoder representations from transformers (VTE-BERT) natural language processing model. Journal of Thrombosis and Haemostasis, 24(7), 2522-2531. https://doi.org/10.1016/j.jtha.2025.07.021
Vancouver
Jafari O, Ma S, Lam BD, Jiang JY, Zhou E, Ranjan M, et al. Development and validation of venous thromboembolism–bidirectional encoder representations from transformers (VTE-BERT) natural language processing model. Journal of Thrombosis and Haemostasis. 2026;24(7):2522-2531. doi:10.1016/j.jtha.2025.07.021
BibTeX
@article{Jafari2026,
author = {Jafari, Omid and Ma, Shengling and Lam, Barbara D and Jiang, Jun Y and Zhou, Emily and Ranjan, Mrinal and Ryu, Justine and Bandyo, Raka and Maghsoudi, Arash and Peng, Bo and Amos, Christopher I and Oluyomi, Abiodun and Fillmore, Nathanael R and La, Jennifer and Li, Ang},
title = {Development and validation of venous thromboembolism–bidirectional encoder representations from transformers (VTE-BERT) natural language processing model},
journal = {Journal of Thrombosis and Haemostasis},
year = {2026},
volume = {24},
number = {7},
pages = {2522--2531},
doi = {10.1016/j.jtha.2025.07.021}
}
EUVIMED