Creating knowledge repositories from biomedical reports: the MEDSYNDIKATE text mining system

Area of Science:

• Medical Informatics
• Natural Language Processing
• Knowledge Representation

Background:

• Medical finding reports contain valuable unstructured data.
• Automated knowledge acquisition from clinical text is challenging.
• Existing systems may lack robust handling of cohesive medical texts.

Purpose of the Study:

• To introduce MEDSYNDIKATE, a natural language processor for knowledge extraction from medical reports.
• To describe a system architecture that handles both single and linked sentences in medical texts.
• To present approaches for ontology engineering to support knowledge acquisition.

Main Methods:

• Developing a natural language processor (MEDSYNDIKATE) for medical text analysis.
• Implementing a system architecture integrating sentence and cohesive text analysis.
• Utilizing two alternative approaches for semi-automatic ontology engineering.
• Evaluating knowledge extraction performance on major syntactic patterns.

Main Results:

• MEDSYNDIKATE successfully transfers medical report content into formal knowledge structures.
• The system architecture accommodates complex linguistic features of medical documents.
• Performance data demonstrates the system's capability in knowledge extraction.
• Ontology engineering methods provide necessary knowledge sources.

Conclusions:

• MEDSYNDIKATE offers an effective solution for automated knowledge acquisition from medical findings.
• The system's architecture and ontology support facilitate the creation of medical knowledge bases.
• The presented performance data validates the system's utility in processing clinical text.