A multi-part matching strategy for mapping LOINC with laboratory terminologies.

Area of Science:

• Medical Informatics
• Computational Linguistics
• Health Data Standards

Background:

• Local laboratory terminologies present challenges for standardized data exchange.
• Mapping these terms to a universal standard like Logical Observation Identifiers Names and Codes (LOINC) is crucial for interoperability.
• Existing ontology matching algorithms require optimization for accuracy and efficiency.

Purpose of the Study:

• To develop and evaluate novel strategies for mapping local laboratory terminologies to LOINC.
• To investigate the impact of term combination probabilities on matching accuracy.
• To reduce manual effort in the LOINC mapping process.

Main Methods:

• Proposed two ontology matching strategies: full name and multi-part.
• The multi-part strategy incorporates the occurrence probability of combined concept parts.
• Validated strategies on three Taiwanese hospital laboratory databases and compared with RELMA's Lab Auto Mapper (LAM).

Main Results:

• The multi-part strategy achieved superior match quality, with F-measure values ranging from 89% to 96%.
• Successfully mapped 70-85% of local terms to LOINC automatically.
• Recommendation feature proposed mappings for an additional 9-20% of terms, with an overall accuracy of 91%.

Conclusions:

• The multi-part strategy significantly outperforms LAM in mapping quality.
• Enables domain experts to perform LOINC matching with reduced manual intervention.
• Term combination probabilities enhance match quality, aid in proposing new LOINC concepts, and decrease processing time.