Country / Region
APAC
Tags
Implementation, Mapping, Tooling
This project focuses on the principles of mapping clinical terms to standard terminology for a multi-institutional research platform. A hybrid mapping approach, combining automated and manual methods, ensures accurate and consistent terminology alignment. Strict mapping guidelines and 1:1 mapping principles are implemented to address terminology variations and data format differences.
This study highlights the critical importance of precise terminology mapping for reliable multi-institutional research.
Description
Key aspects include: Hybrid Mapping Process: Combines automated and manual methods to ensure semantic consistency and compliance with global standards. * Validation & Expert Involvement: Cross-validation and expert consultations resolve mapping discrepancies and inactive concept issues. Mapping Strategies: Establishes mapping principles for LOINC, RxNorm, and SNOMED CT by identifying the characteristics of each terminology and analyzing the current state of local terms, in order to develop practically applicable mapping guidelines. Implementation & Outcomes: Enhances mapping accuracy, improves cohort generation, and facilitates multi-institutional research through a user-friendly platform with advanced code search capabilities. Challenges & Future Directions: Addresses inter-institutional inconsistencies and SNOMED CT limitations while highlighting the need for trained terminology specialists in Korea. Future improvements include advanced search options and attribute-based retrieval for better usability. Ultimately, this study provides a scalable strategy for clinical terminology mapping, fostering efficient multi-institutional research and supporting global healthcare innovation. Accuracy, recall, and other performance metrics are used to compare the proposed method with traditional rule-based standardization approaches.
Scope
SNOMED CT is utilized in this study as the primary standard terminology system for mapping clinical terms, particularly for diagnosis, chief complaints, treatment procedures, measurements, and medical history. It is employed to ensure semantic consistency and interoperability across institutions.
Key Usage of SNOMED CT in the Mapping Process:
1. Standardization & Interoperability:
* SNOMED CT is selected for domains where no widely accepted terminology exists due to its robust modeling capabilities.
* It provides a structured approach to ensure semantic equivalence in mapped clinical terms.
2. Hybrid Mapping Approach (Automated + Manual):
* Automated Mapping:
* Chipmunk, Kakao Healthcare’s in-house terminology tool, uses vector similarity to map local terms to SNOMED CT.
* Manual Mapping:
* Clinical experts review and correct automated mappings, ensuring accuracy and consistency.
* Terms without direct SNOMED CT matches undergo post-coordination, where new concepts are created.
3. New Concept authoring & Validation:
* If pre-coordinated concepts are unavailable, Chipmunk supports the post-coordination process to generate new SNOMED CT concepts.
* Attributes and values are refined within their domains, and validation is performed using the Machine Readable Concept Model (MRCM).
* Concepts are assigned a Kakao Healthcare extension namespace (1000305) and updated biannually.
4. Mapping Guidelines & Constraints:
* SNOMED CT Editorial Guidelines are followed, but internal rules are created for cases not covered.
* 1:1 mapping is preferred to ensure data consistency in multi-institutional research.
* Uses domain-specific semantic restrictions, attribute prioritization for post-coordination, and tailored mapping approaches.
* Many-to-one mapping is allowed for refined local codes (e.g., multiple diagnosis terms mapped to a single SNOMED CT concept).
By implementing SNOMED CT in this way, the study ensures semantic integrity, interoperability, and reliable data alignment across institutions, ultimately supporting multi-institutional research and healthcare innovation.
How SNOMED CT will be used
SNOMED CT was chosen as the primary standard terminology system in this study due to its comprehensive coverage, structured modeling capabilities, and global acceptance. The selection was driven by several key factors:
* SNOMED CT provides a broad and detailed representation of clinical concepts, covering diagnoses, chief complaints, procedures, medical history, and measurements.
* Multi-institutional research requires a standardized terminology that enables consistent data integration across different medical centers.
* Unlike some terminologies with limited predefined terms, SNOMED CT allows post-coordination, enabling the creation of new concepts when necessary. This flexibility was crucial for ensuring the granularity needed in multi-institutional research
* SNOMED CT is widely recognized and used internationally, making it an ideal choice for a research platform aiming for global compatibility.
* Using SNOMED CT enhances data retrieval and searchability, facilitating efficient cohort identification for research.
* The medical field evolves rapidly, and SNOMED CT provides a regular update mechanism, ensuring that new concepts can be added as medical knowledge advances.
SNOMED CT was selected for its robustness, flexibility, and ability to ensure semantic consistency across institutions. Its structured approach to concept representation, interoperability, and scalability makes it an essential component of the HRS (Healthcare data Research Suite) platform and UDM model (Universal Data Model by Kakao Healthcare), supporting efficient multi-institutional research and healthcare innovation.
Why SNOMED CT will be used
Contact
