New natural language processing tool available to investigators

CTSI’s Informatics Consulting Services (ICS) recently began offering a new self-service tool to University investigators that retrieves unstructured clinical notes from the electronic health record.

The Natural Language Processing - Patient Information Extraction for Research (NLP-PIER) tool can be used for biomedical research, including recruitment planning, retrospective cohort studies, and observational studies. It includes capabilities to filter and search documents using HL7-LOINC Document Ontology features, search via keywords and Unified Medical Language System (UMLS) concept and semantic type, and perform Boolean queries. 

Records in the tool’s initial release include:

  • Fairview Health Services inpatient and outpatient clinical notes in the Fairview Epic electronic health record, including: 
    • Outpatient documents from more than 40 clinics from 2005 to present.
    • Inpatient documents from Fairview hospitals, including University of Minnesota Medical Center, from 2011 to present.
  • Outpatient documents from University of Minnesota Physicians from 2011 to present.

The repository, which is updated on a weekly basis to include new notes and latest versions to existing notes, uses Elasticsearch technology and features the University of Minnesota Clinical NLP/IE program open source NLP application, the BioMedical Information Collection and Understanding System. 

The tool is a collaborative effort between CTSI, the NLP/IE program, the Minnesota Supercomputing Institute (MSI), and the Academic Health Center’s Research Development and Support (RDS) group. 

The PIER information extraction tool is available to University biomedical and clinical researchers through the CTSI ICS and can be requested through the Clinical Data Request form