Programmatic Theme: Informatics Implementation
Abstract: Data from medical examiner offices are not commonly used in informatics but may contain information not in medical records. However, the vast majority of data is not standardized and is available only in large free text fields. We sought to extract information from the medical examiner database using Canary, a natural language processing tool. The text was then standardized to fit the selected normative answer list for each field. Multiple terminology and vocabulary standards from a variety of settings were utilized as data came from the medical examiner and interviews with next of kin. Thirty-seven percent of the metadata fields could be mapped directly to existing standards, twenty-fivepercent required a modification, and thirty-eight required creation of a standardized normative answer list. The newly formed database (New Mexico Decedent Image Database (NMDID)), will be available to researchers and educators at the beginning of 2020.
Learning Objective: By the end of the talk the audience will know:
1) the steps for determing vocabulary standards for new databases.
2) how Canary extracts data from large free text fields.
Shamsi Berry (Presenter)
University of Mississippi Medical Center
Heather Edgar, University of New Mexico