GALEN based formal representation of ICD10.

Area of Science:

• Medical Informatics
• Ontology Engineering
• Computational Linguistics

Background:

• The International Classification of Diseases (ICD10) is crucial for healthcare data but lacks formal representation, hindering computer-assisted coding.
• Existing representation languages for medical ontologies have limitations, necessitating transformation to standard formats like OWL.
• Decidability is a key challenge in developing effective computer-assisted coding systems.

Purpose of the Study:

• To create a knowledge-intensive coding support tool for ICD10 using a formal ontology.
• To represent ICD10 categories in description logic for decidability and reusability.
• To develop a test system for verifying the feasibility of the ontology-based approach.

Main Methods:

• Representing ICD10 category semantics using the GALEN Core Reference Model.
• Transforming the ontology from GRAIL to OWL DL due to GRAIL's deficiencies.
• Implementing a Prolog-based test system to extract and classify disease concepts.

Main Results:

• Formal representation of the first two ICD10 chapters (infectious diseases and neoplasms) is nearly complete.
• The ontology was successfully converted to OWL DL.
• The test system achieved 84% recall but only 45% precision in identifying gastrointestinal oncology diseases.

Conclusions:

• The formal ontology approach is feasible for developing ICD10 coding support tools.
• Further development is needed for the classifier module to improve precision.
• Future work will incorporate the Foundational Model of Anatomy (FMA) as an anatomical reference ontology.