Clinical Terminology Mapping Service Based on Information Retrieval.

Area of Science:

• Medical Informatics
• Natural Language Processing
• Clinical Terminology

Background:

• Standardized clinical terminology is crucial for semantic interoperability in healthcare.
• Manual mapping of local terms to standards like SNOMED CT is labor-intensive and requires significant expertise.
• Existing methods face challenges in accuracy and efficiency.

Purpose of the Study:

• To develop and evaluate an information retrieval (IR) method for mapping clinical terms to SNOMED CT concept descriptions.
• To leverage rich synonyms and edge n-gram indexing to improve mapping accuracy.
• To provide a free mapping support service to assist terminology experts.

Main Methods:

• Developed an IR approach using Elasticsearch for indexing and query processing.
• Utilized edge n-grams and an enriched synonym database from the SPECIALIST Lexicon.
• Tested four optimized indexes individually and in combination on 1,753 ICD-9-CM to SNOMED CT mappings.
• Compared the developed method against MetaMap for performance evaluation.

Main Results:

• The combination of rich synonyms and edge n-gram indexing significantly improved mapping accuracy.
• Indexes incorporating both techniques outperformed those using either alone.
• The proposed method demonstrated superior performance compared to MetaMap.
• Enhanced capability in handling complex medical terminology and improving mapping quality.

Conclusions:

• An effective IR method using rich synonyms and edge n-grams for clinical term to SNOMED CT mapping was introduced.
• The approach shows promise in reducing manual mapping efforts and improving accuracy.
• Future work will focus on further enhancing accuracy and result evaluation methods.