HSR&D Citation Abstract
Search | Search by Center | Search by Source | Keywords in Title
Information Extraction and Visualization Toolkit (IE-VIZ) Pipeline
Chapman WW. Information Extraction and Visualization Toolkit (IE-VIZ) Pipeline. 2017 Oct 1.
Brief Description: This informatics tool is being validated and will advance visualization workbenches and create effective clinical data visualizations. The foundation of the infrastructure is a pair of ontologies that model entities and events in clinical text and their modifiers. Use of the ontologies allows us to capture explicit case/phenotype definitions from experts, to share knowledge, and to apply more than one NLP tool to the extraction problems. The infrastructure is novel and enables stronger partnership between domain experts and Natural Language Processing (NLP) developers.
Related Products: There are multiple dashboards now being developed for the EHR. The first such system, called NobleMentions, was developed at the University of Pittsburg and is being integrated into IE-Viz. A second NLP system, called Moonstone, which was developed at the University of Utah, has also been added to IE-Viz.
Adoption by VA Facilities: Application for select conditions and widespread validation is underway.
New Research or Operations Funding to Further Develop, Test or Disseminate: CREATE funding allows for continuation of large scale validation.
Led to or Shaped VA National Program or Clinical Practice(s): Dashboards assist clinicians with rapid and more accurate decision making than lengthy text notes. IE-Viz system allows advanced users to build and integrate their own NLP systems.
Advances in FY 17: The command line interface for IE-Viz was created and is being tested. Several changes were made to the IE-Viz data model based on work creating a cancer phenotyping natural language processing pipeline developed at the University of Pittsburg. Also, we are adding new features to facilitate ease of use and long term maintainability.