Standardized Representation of Clinical Study Data Dictionaries with CIMI Archetypes

Area of Science:

• Clinical Informatics
• Biomedical Data Science
• Health Data Standardization

Background:

• Clinical study data is commonly presented in tabular formats.
• Lack of standardized metadata elements in data dictionaries hinders study harmonization and interoperability.
• Existing data dictionaries lack a common format for representation.

Purpose of the Study:

• To propose standardized archetypes for representing data dictionaries.
• To overcome challenges in data harmonization and interoperability.
• To establish a common format for data dictionary models.

Main Methods:

• Representing data dictionaries using standardized archetypes.
• Utilizing Archetype Modeling Language (AML) from the Clinical Information Modeling Initiative (CIMI).
• Mapping three data dictionaries (dbGAP, PheKB, TCGA) to AML archetypes by aligning variable definitions.

Main Results:

• Achieved near-complete alignment of data dictionaries with AML archetypes.
• Successfully mapped data dictionaries into valid AML models.
• Captured all metadata elements of the data dictionary models.

Conclusions:

• Standardized archetypes in AML can effectively represent data dictionaries.
• This approach facilitates harmonization of data models for improved quality and semantic interoperability.
• Enables better downstream data integration for researchers and subject matter experts.