Artificial Intelligence-based Automated International Classification of Diseases Coding: A Systematic Review.

Area of Science:

• Health Informatics
• Artificial Intelligence in Healthcare
• Clinical Coding Systems

Background:

• Automated clinical coding using AI is crucial for healthcare efficiency.
• International Classification of Diseases (ICD) coding accuracy is vital for billing and research.
• Existing manual coding processes face challenges in speed and consistency.

Purpose of the Study:

• To systematically review current knowledge on AI-based automated ICD coding.
• To identify and synthesize the challenges, benefits, and future research directions in this field.
• To provide insights for advancing automated ICD coding systems.

Main Methods:

• Systematic literature search conducted on January 1, 2024, across major databases (PubMed, Embase, Scopus, Web of Science).
• Adherence to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.
• Inclusion of studies focusing on AI-driven ICD coding challenges, advantages, and research gaps; analysis of 8 selected studies out of 12,641 records.

Main Results:

• Six key challenges identified: extensive label space, imbalanced data, lengthy documents, interpretability, ethical concerns, and lack of transparency.
• Ten major benefits highlighted: improved decision-making, data standardization, and enhanced coding accuracy.
• Eight future research directions proposed: interdisciplinary collaboration, transfer learning, transparency, and active learning.

Conclusions:

• AI-based automated ICD coding presents significant potential to revolutionize clinical coding.
• Addressing identified challenges is crucial for successful implementation.
• Further research in areas like transfer learning and transparency is recommended to optimize AI-driven coding systems.