BGLM

BGLM maps local laboratory test codes from electronic health records (EHR) to Logical Observation Identifiers Names and Codes (LOINC) using distributional similarity on large patient datasets to harmonize laboratory data for interoperability.

Key Features:

• Data-Driven Mapping: Uses extensive patient data stored in EHR systems rather than relying solely on text-mining to derive mappings between local codes and LOINC.
• Distributional Similarity Approach: Employs distributional similarity analyses of test result patterns to identify LOINC matches for local laboratory codes.
• Robust Multi-Language Support: Performs code mapping independent of language, enabling application across diverse linguistic settings.
• Performance Validation: Validated on real-world datasets with high mapping precision reported when appropriate false discovery rate controls are applied.
• Integration with Existing Tools: Produces results that can enhance performance of the Regenstrief LOINC Mapping Assistant (RELMA) for LOINC mapping workflows.

Scientific Applications:

• EHR Harmonization for Multi-Institutional Studies: Standardizes laboratory test codes across institutions to enable pooled analyses of clinical data.
• International Health Initiatives: Facilitates cross-border data integration by mapping local codes to LOINC regardless of language differences.
• Large-Scale Clinical Research: Supports aggregation and comparative analysis of laboratory results across large patient cohorts.
• Personalized Medicine Research: Enables consistent laboratory data representation to inform individualized clinical and translational studies.

Methodology:

Applies big data analytics to analyze distributional similarities of laboratory test result patterns in EHR patient data to identify probable LOINC matches and employs false discovery rate controls to ensure mapping precision.