Automated identification of fall-related injuries in unstructured clinical notes
American Journal of Epidemiology · 2025
DOI: 10.1093/aje/kwae240 ↗Kurz gefasst
Fachartikel in der Zeitschrift American Journal of Epidemiology, 2025. Verfasst von Wendong Ge und 10 weiteren Autor:innen. Beteiligte Einrichtung: Department of Neurology, Massachusetts General Hospital, Boston, MA 02114, United States. Hauptthemen laut MEDLINE: Accidental Falls, Natural Language Processing, Electronic Health Records, Wounds and Injuries. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 41 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NINDS NIH HHS, NIH, NHLBI NIH HHS.
Abstract (Originalsprache)
Fall-related injuries (FRIs) are a major cause of hospitalizations among older patients, but identifying them in unstructured clinical notes poses challenges for large-scale research. In this study, we developed and evaluated natural language processing (NLP) models to address this issue. We utilized all available clinical notes from the Mass General Brigham health-care system for 2100 older adults, identifying 154 949 paragraphs of interest through automatic scanning for FRI-related keywords. Two clinical experts directly labeled 5000 paragraphs to generate benchmark-standard labels, while 3689 validated patterns were annotated, indirectly labeling 93 157 paragraphs as validated-standard labels. Five NLP models, including vanilla bidirectional encoder representations from transformers (BERT), the robustly optimized BERT approach (RoBERTa), ClinicalBERT, DistilBERT, and support vector machine (SVM), were trained using 2000 benchmark paragraphs and all validated paragraphs. BERT-based models were trained in 3 stages: masked language modeling, general boolean question-answering, and question-answering for FRIs. For validation, 500 benchmark paragraphs were used, and the remaining 2500 were used for testing. Performance metrics (precision, recall, F1 scores, area under the receiver operating characteristic curve [AUROC], and area under the precision-recall [AUPR] curve) were employed by comparison, with RoBERTa showing the best performance. Precision was 0.90 (95% CI, 0.88-0.91), recall was 0.91 (95% CI, 0.90-0.93), the F1 score was 0.91 (95% CI, 0.89-0.92), and the AUROC and AUPR curves were [both??] 0.96 (95% CI, 0.95-0.97). These NLP models accurately identify FRIs from unstructured clinical notes, potentially enhancing clinical-notes–based research efficiency.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Neurology, Massachusetts General Hospital, Boston, MA 02114, United States.
- Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA 02115, United States.
- Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA 02115, United States.
- Department of Psychiatry, Massachusetts General Hospital, Boston, MA 02114, United States.
- Department of Health Care Policy, Harvard Medical School, Boston, MA 02115, United States.
- Mongan Institute, Massachusetts General Hospital, Boston, MA 02114, United States.
Förderung
- NINDS NIH HHS
- NIH
- NHLBI NIH HHS
- Epilepsy Foundation
Zitieren
APA
Ge, W., Godeiro Coelho, L. M., Donahue, M. A., Rice, H. J., Blacker, D., Hsu, J., Newhouse, J. P., Hernández-Díaz, S., Haneuse, S., Westover, B., & Moura, L. M. V. R. (2025). Automated identification of fall-related injuries in unstructured clinical notes. American Journal of Epidemiology. https://doi.org/10.1093/aje/kwae240
Vancouver
Ge W, Godeiro Coelho LM, Donahue MA, Rice HJ, Blacker D, Hsu J, et al. Automated identification of fall-related injuries in unstructured clinical notes. American Journal of Epidemiology. 2025. doi:10.1093/aje/kwae240
BibTeX
@article{Ge2025,
author = {Ge, Wendong and Godeiro Coelho, Lilian M and Donahue, Maria A and Rice, Hunter J and Blacker, Deborah and Hsu, John and Newhouse, Joseph P and Hernández-Díaz, Sonia and Haneuse, Sebastien and Westover, Brandon and Moura, Lidia M V R},
title = {Automated identification of fall-related injuries in unstructured clinical notes},
journal = {American Journal of Epidemiology},
year = {2025},
doi = {10.1093/aje/kwae240}
}
EUVIMED