Leveraging Semantics in WordNet to Facilitate the Computer-Assisted Coding of ICD-11.

Area of Science:

• Medical Informatics
• Computational Linguistics
• Health Information Management

Background:

• The International Classification of Diseases (ICD) is crucial for health statistics and disease overview.
• Accurate ICD coding is essential for reliable health data and research.
• Existing computer-assisted coding for ICD-11 has limitations in precision and recall.

Purpose of the Study:

• To develop an improved computer-assisted coding framework for ICD-11.
• To leverage ICD-11 data structures and WordNet semantics for enhanced coding.
• To increase the accuracy and efficiency of ICD-11 coding processes.

Main Methods:

• Proposed a computer-assisted coding framework integrating WordNet and the ICD-11 API.
• Developed a network (CodeNet) combining ICD-11 entity relations and WordNet synonym sets.
• Designed an algorithm to generate ICD-11 code candidates using CodeNet and disease-related text.

Main Results:

• Achieved a precision of 84% and recall of 89%, outperforming the existing ICD-11 API (65% precision, 81% recall).
• Reduced the ICD-11 coding failure rate by up to 8% compared to other methods.
• Demonstrated the effectiveness of proposed similarity and percentage thresholds for performance tuning.

Conclusions:

• Integrating WordNet semantics with ICD-11 structures enhances computer-aided coding reliability.
• The developed method offers a more accurate and efficient solution for ICD-11 coders.
• This approach has the potential to improve the overall quality of health statistics and disease surveillance.