Related Experiment Video
Updated: Nov 11, 2025
A Metadata Extraction Approach for Clinical Case Reports to Enable Advanced Understanding of Biomedical Concepts
Published on: September 20, 2018
Extracting COVID-19 diagnoses and symptoms from clinical text: A new annotated corpus and neural event extraction
Kevin Lybarger1, Mari Ostendorf2, Matthew Thompson3
1Biomedical & Health Informatics, University of Washington, Box 358047, Seattle, WA 98109, USA.
Insights
This study introduces the CACT Corpus and a novel event extraction model to analyze free-text clinical notes for COVID-19 research. The model accurately identifies COVID-19 diagnoses and symptoms, improving prediction of test results.
Area of Science:
- Medical Informatics
- Natural Language Processing
- Public Health
Background:
- The COVID-19 pandemic necessitates advanced methods for analyzing clinical data to understand disease spread, symptoms, and healthcare needs.
- Free-text clinical notes are rich in information but require automated extraction for large-scale studies.
- Existing methods may not fully capture the nuances of symptomology and assertion in clinical text.
Purpose of the Study:
- To develop and evaluate a novel span-based event extraction model for analyzing COVID-19-related information in clinical notes.
- To introduce the COVID-19 Annotated Clinical Text (CACT) Corpus for facilitating research on COVID-19.
- To assess the utility of automatically extracted symptom information in predicting COVID-19 test results.
Main Methods:
- Creation of the CACT Corpus with 1,472 annotated clinical notes detailing COVID-19 diagnoses, testing, and presentation.
- Development of a span-based event extraction model to jointly identify COVID-19 and symptom events with assertion values.
- Comparison of the developed model against a MetaMapLite-based extractor for symptom identification.
- Secondary application: Prediction of COVID-19 test results using structured data combined with automatically extracted symptom information.
Main Results:
- The span-based event extraction model achieved high performance (0.83-0.97 F1 for events, 0.73-0.79 F1 for assertions) in identifying COVID-19 and symptom events.
- The developed model outperformed MetaMapLite in extracting symptoms with assertion values.
- Incorporating automatically extracted symptoms improved the prediction of COVID-19 test results compared to using structured data alone.
Conclusions:
- The CACT Corpus and the proposed span-based event extraction model provide valuable tools for COVID-19 research using clinical text.
- Automated extraction of clinical information from free-text notes can significantly enhance our understanding of COVID-19 and improve predictive models.
- This approach holds promise for advancing public health surveillance and clinical decision-making during pandemics.
Abstract:
Coronavirus disease 2019 (COVID-19) is a global pandemic. Although much has been learned about the novel coronavirus since its emergence, there are many open questions related to tracking its spread, describing symptomology, predicting the severity of infection, and forecasting healthcare utilization. Free-text clinical notes contain critical information for resolving these questions. Data-driven, automatic information extraction models are needed to use this text-encoded information in large-scale studies. This work presents a new clinical corpus, referred to as the COVID-19 Annotated Clinical Text (CACT) Corpus, which comprises 1,472 notes with detailed annotations characterizing COVID-19 diagnoses, testing, and clinical presentation. We introduce a span-based event extraction model that jointly extracts all annotated phenomena, achieving high performance in identifying COVID-19 and symptom events with associated assertion values (0.83-0.97 F1 for events and 0.73-0.79 F1 for assertions). Our span-based event extraction model outperforms an extractor built on MetaMapLite for the identification of symptoms with assertion values. In a secondary use application, we predicted COVID-19 test results using structured patient data (e.g. vital signs and laboratory results) and automatically extracted symptom information, to explore the clinical presentation of COVID-19. Automatically extracted symptoms improve COVID-19 prediction performance, beyond structured data alone.
