Talk to the Veterans Crisis Line now
An official website of the United States government
Search | Search by Center | Search by Source | Keywords in Title
Gundlapalli AV, Samore MH, Palmer M, Tuteja AK, Carter M, Shen S, South B, Forbush T, Divita G. Annotation of Symptoms in VA Clinical Documents. Poster session presented at: Integrating Data for Analysis, Anonymization, and Sharing Annual Conference; 2012 Sep 29; La Jolla, California.
Background: - 2+ million troops have been deployed in support of Operation Enduring Freedom (OEF), Operation Iraqi Freedom (OIF) - Medically unexplained syndromes (MUS): conditions diagnosed on the basis of symptom constellations and characterized by a lack of well-defined pathogenesis (e.g. Irritable bowel syndrome (IBS), chronic fatigue syndrome (CFS), Fibromyalgia) - Associations between deployment and MUS have been demonstrated by epidemiologic studies of Veterans from other conflicts - Little is known about MUS in troops deployed to support OEF/OIF - Symptoms were defined broadly as any feature of disease noticed or experienced by the patient Objectives: - Develop a manually annotated reference standard of positively asserted symptoms expressed in VA clinical documents Methods Random sample of clinical encounter documents from a national cohort of OEF/OIF Veterans - Guidelines were created iteratively using clinician input and review - Schema were created to match the guidelines - Annotations done using Prot g 3.3.1 and Knowtator version 1.9 Beta 2 - Trained 2 annotators, 1 adjudicator, 1 reviewer - Achieved acceptable inter annotator agreement (IAA) using synthetic and real documents - Batches of 10 documents with every fifth batch being reviewed for quality control for first 40 batches - Remaining batches focused on volume and used a single annotator Results - 5,572 symptoms identified - 543 marked as subjective mentions - Lexicon of 2,384 unique symptoms created - Additional contextual data was collected about the symptoms - Change in status (371) - Temporality (566) - Average IAA was 58.71% Discussion: - Identifying symptoms is a difficult task that involves both information extraction and classification steps Future Steps: - Use the reference standard to train and evaluate an automated NLP system - Second round of annotation and review will be focused on symptom causality and explainability - Characterize the occurrence of MUS OEF/OIF Veterans - Provide cohorts of Veterans with MUS for new studies and clinical trials Impact Using an automated NLP system to identify and examine symptom clusters will provide data that can be used to continually assess the health status of OEF/OIF Veterans. The ability to measure the burden of MUS has the potential to provide more comprehensive health care services to all Veterans. Acknowledgements -Project Funded by Department of Veterans Affairs, Office of Research and Development, Health Services Research and Development Project # HIR 10-001 -This project has used data resources from VA Informatics and Computing Infrastructure (VINCI) -Contact : Miland Palmer, Research Project Manager, Department of Veterans Affairs Medical Center, Salt Lake City, UT Miland.Palmer@va.gov -The views expressed are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government
