Annotation methods to develop and evaluate an expert system based on natural language processing in electronic...

Area of Science:

• Health Informatics
• Medical Informatics
• Epidemiology

Background:

• Generating meaningful epidemiological data from unstructured French medical records presents significant challenges.
• Existing methods for secondary use of electronic health records (EHR) data are often domain-specific and lack interoperability.

Purpose of the Study:

• To develop a generic IT solution for extracting epidemiological data from French medical records.
• To create a care pathway-oriented conceptual model and annotation method for building expert system knowledge bases.
• To enable the secondary use of EHR data through a standardized factual database.

Main Methods:

• Developed a system combining a medical terminology server, semantic analyser, and knowledge base.
• Implemented a semi-automatic annotation method using the MedIndex software.
• Utilized a cross-lingual multi-termino-ontology portal for medical concept coding and temporal labeling.
• Focused on a care pathway conceptual model for annotating medical events.

Main Results:

• A generic IT solution was developed, integrating key components for data extraction.
• A novel annotation method was established, facilitating semi-automatic coding and temporal labeling of medical concepts.
• The system facilitates the creation of a structured database from textual medical records.

Conclusions:

• The developed system and annotation method provide a generic approach for generating epidemiological data from French EHRs.
• This facilitates the secondary use of medical records, enhancing epidemiological research.
• The care pathway model and ontology-driven annotation contribute to a standardized factual database.