Explainable AI for Medical Coding
Reframing ICD coding as an entity linking challenge for transparency
This research transforms clinical coding from a traditional classification task into an entity linking problem, making the process more transparent and traceable for healthcare professionals.
- Provides explicit evidence for each code by linking it to specific text passages in medical records
- Addresses a critical need for justifiable coding decisions in healthcare environments
- Creates a more auditable and trustworthy system for automated medical coding
- Enables medical coders to verify coding decisions against source documentation
This approach matters for healthcare organizations by reducing manual review time, improving regulatory compliance, and increasing coder confidence in AI-assisted processes.