2011 HSR&D National Meeting Abstract
2004 — An Introduction to Natural Language Processing in the VA
DuVall SL (VA Salt Lake City Health Care System), South BR
(VA Salt Lake City Health Care System), Meystre SM
(University of Utah), D'Avolio LW
(Massachusetts Veterans Epidemiology Research and Information Center)
The majority of electronic clinical documentation in the VA is stored as “free text” in clinical notes rather than as structured, coded data. An advantage of free text is that it gives clinical authors autonomy in expressing their thoughts. The variety of ways used to express information in text means that although this data is rich and descriptive, it is locked away, unable to be used in computerized research and decision support. Natural language processing (NLP) is key to unlocking the concepts, context, and relationships found in these notes. While NLP is not a “solved” science, there are many tasks that NLP can do reliably. Extracting concepts (symptoms, diseases, medications) and values (ejection fraction value, lab values, vital signs) that are stored in the text is one example. More complex tasks, such as determining what caused an event of interest or why a patient discontinued a medication can also be addressed using the right tools. More than one billion text notes are stored in the VA with 600,000 new notes created every day. Researchers can access these text notes in approved studies through the VA Informatics and Research Infrastructure (VINCI). This workshop will introduce researchers to NLP and explore ways that NLP can support ongoing research.
Discussion of what NLP is and different approaches for processing clinical texts;
explanation of the role that NLP can play in health services research;
interactive demonstration of the annotation process used to create reference standards and the tools used; and interactive demonstration of the Automated Retrieval Console (ARC) NLP tool.
Health services researchers, epidemiologists, and statisticians interested in learning the ways that NLP can be used to enhance research endeavors inside the VA. The focus will be on the methods and tools that exist and are ready to use in VINCI.
Assumed Audience Familiarity with Topic:
We anticipate workshop participants will have cursory knowledge of NLP and feel that some aspect of NLP could address a current or future research need.