Development of structured ICD-10 and its application to computer-assisted ICD coding.

Area of Science:

• Medical Informatics
• Ontology Engineering
• Health Information Systems

Background:

• Accurate medical coding is crucial for healthcare data management and billing.
• Existing ICD-10 coding systems face challenges in handling complex semantic relationships and natural language input.
• The need for standardized, machine-readable representations of medical classification systems is increasing.

Purpose of the Study:

• To develop a formal representation framework for ICD-10.
• To create a methodology for computer-assisted ICD coding using this formal representation.
• To inform the development of information models for the upcoming ICD-11 revision.

Main Methods:

• Analysis and structuring of ICD-10 category meanings across 15 chapters.
• Expansion of structured ICD-10 (S-ICD10) with Japanese Standard Disease Names.
• Refinement of an information model with 74 semantic link types.
• Development of an ICD coding module and a 'Coding Principle'.

Main Results:

• A formal representation framework for ICD-10 was established.
• The structured ICD-10 (S-ICD10) was expanded with additional semantic details.
• The computer-assisted ICD coding module achieved over 70% accuracy for four ICD-10 chapters.
• The feasibility of the proposed coding framework was demonstrated.

Conclusions:

• The developed framework provides a bridge between ontological information and natural language for ICD-10.
• The methodology enables accurate computer-assisted ICD coding.
• The findings support the potential for enhancing the ICD-11 revision with improved information models.