LLM-Assisted Tool Streamlines SNOMED CT Mapping
AI-assisted, source-derived brief produced by the Let's Data Science Automated News Desk. The source material used is linked on this page.
- Source event:
- first reported
- LDS brief:
- publication time is not available in the public LDS lifecycle record
Researchers at Kakao Healthcare and collaborators published in JMIR Medical Informatics 2026 describe an LLM-assisted tool using GPT-4o to automate SNOMED CT mapping and new-concept authoring across nine South Korean university hospitals. The system achieved top-5 diagnostic mapping accuracies ranging from 89.7% to 98.7%, cut manual mapping rates by 30% and overall workload by up to 90%. Time to map and create concepts fell ~75%, with duplicates down 83% and rule violations down 72%.
Key Points
- 1Achieved high mapping accuracy: top-5 diagnostic accuracy 98.7%, 89.7%, 98.5%, 92.8% across four institutions
- 2Reduced manual effort substantially, cutting manual mapping rates by 30% and overall workload by up to 90%
- 3Enabled faster, higher-quality concept authoring: 75% faster creation, 83% fewer duplicates, 72% fewer modeling violations
Scoring Rationale
Demonstrated robust, multi-institutional LLM mapping with strong quantitative gains; limited by integration and translation dependencies.
Sources
Public references used for this report.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems