Leveraging Shannon Entropy to Validate the Transition between ICD-10 and ICD-11

Area of Science:

• Health Informatics
• Medical Coding Systems
• Information Theory

Background:

• The transition from International Classification of Diseases 10th revision (ICD-10) to ICD-11 necessitates robust validation methods.
• Existing methods may not adequately capture information changes during this complex coding system transition.

Purpose of the Study:

• To propose a novel mapping framework utilizing entropy-based metrics for validating the transition from ICD-10 to ICD-11.
• To evaluate information changes from single-code, single-disease, and multiple-disease perspectives.

Main Methods:

• Development of a framework using ICD-11 tabular lists and mapping tables.
• Application of Shannon entropy to derive validation metrics: standardizing rate (SR), uncertainty rate (UR), and information gain (IG).
• Validation using a large-scale ICD-10 coded dataset (377,589 records).

Main Results:

• The proposed metrics simplify the evaluation of the ICD-10 to ICD-11 transition.
• Standardizing Rate (SR) indicated ~60% of ICD-10 codes lacked standard WHO mapping.
• Uncertainty Rate (UR) achieved 86.21% precise mapping accuracy.
• Information Gain (IG) showed ~57% of records experienced increased uncertainty post-transition.

Conclusions:

• The developed entropy-based metrics offer a reliable and effective approach for validating ICD-10 to ICD-11 code mapping.
• ICD-11's new features facilitate improved mapping reliability.
• The framework highlights areas needing attention for smoother transitions between coding systems.