A natural language understanding system combining syntactic and semantic techniques

Area of Science:

• Medical Informatics
• Natural Language Processing
• Clinical Data Management

Background:

• A significant portion of medical records exists as unstructured free text.
• This free text data is largely inaccessible for automated analysis, hindering clinical decision support and research.
• Computerized medical information systems improve accessibility but not analytical usability of free text reports.

Purpose of the Study:

• To develop and evaluate a Natural Language Understanding System (NLUS) for processing free text chest radiograph reports.
• To extract and structure clinical data from radiograph reports for database storage.
• To enhance the utility of free text medical data for automated applications.

Main Methods:

• Development of an experimental Natural Language Understanding System (NLUS).
• The NLUS is designed to parse free text reports from chest radiographs.
• Extracted clinical data is stored in a structured medical database.

Main Results:

• The experimental NLUS successfully parsed chest radiograph reports.
• Clinical data was effectively extracted and stored in a database.
• Demonstrated the feasibility of encoding free text radiology reports.

Conclusions:

• Natural Language Understanding Systems offer a viable method for accessing unstructured clinical data.
• Automated parsing of chest radiograph reports can make valuable information available for research and decision support.
• This approach can significantly improve the usability of data within computerized medical information systems.