AI Mapping of In-House Codes to LOINC Codes Using Laboratory Test Results Excluding Test Names: Toward International...

Area of Science:

• Medical Informatics
• Computational Linguistics
• Machine Learning

Background:

• Growing need for integrated medical databases across facilities.
• Standardized coding (e.g., LOINC codes) is crucial for data interoperability.
• Challenges in Japanese natural language processing (NLP) for medical terms hinder automated mapping.

Purpose of the Study:

• To develop a novel machine learning-based method for mapping in-house medical codes to LOINC codes.
• To overcome limitations of NLP in Japanese due to scarce medical term corpora.
• To leverage test result values for accurate code mapping, bypassing reliance on test names.

Main Methods:

• Developed a machine learning model for code mapping.
• Utilized test result values as features for the mapping process.
• Avoided direct reliance on test names to circumvent NLP challenges.

Main Results:

• Achieved high mapping accuracy (≥70%) for 80.4% of targeted analytes.
• Demonstrated the effectiveness of the value-leveraging approach.
• Successfully mapped in-house codes to LOINC codes without extensive NLP.

Conclusions:

• The proposed method facilitates easier mapping to standardized LOINC codes, especially in languages with NLP challenges.
• Ensures accurate mapping regardless of the source data language.
• Supports the creation of integrated medical databases across diverse healthcare settings.