Healthcare professionals use 40% of their time to record patient data. Despite large recording time, data is often incomplete and unstructured which leads to frustration.
Recording free text remains attractive due to limited functionality and support of the user interface of EPIC, immature terminological systems and insufficient value of structured recording. Use of free text maintains the administrative burden: data have to be recorded multiple times, and reuse of data, e.g. for research, management information and quality audits is hampered due to lack of data quality and data structure. The aim of this project is to solve this catch-22 by natural language processing to recognize changed diagnoses descriptions causing discrepancies between description and underlying DHD and SNOMED codes; (semi)automated processing of these changes into new codes and/or proposals to update the DHD diagnoses thesaurus; feedback and training to physicians regarding data quality and data reuse.