Quality Assurance of Cancer Study Common Data Elements Using A Post-Coordination Approach.

Area of Science:

• Biomedical Informatics
• Clinical Data Standards
• Cancer Research Data Management

Background:

• Domain-specific common data elements (CDEs) are crucial for standardized clinical research data.
• A significant challenge is the absence of automated quality assurance (QA) tools for CDEs in clinical study domains.
• Developing robust QA tools is essential for improving data accuracy and reliability in cancer research.

Purpose of the Study:

• To prototype and evaluate a novel QA tool for cancer CDEs.
• To implement a post-coordination approach for automated CDE quality checking.
• To enhance the standards-based storage and retrieval of clinical research data.

Main Methods:

• Integrated NCI caDSR CDEs and The Cancer Genome Atlas (TCGA) data dictionaries into a Resource Description Framework (RDF) data store.
• Designed a compositional expression pattern based on ISO/IEC 11179 Data Element Concept model.
• Developed a transformation tool to convert expressions to Web Ontology Language (OWL) syntax, utilizing reasoning and explanation services for QA.

Main Results:

• The developed system successfully identified duplicate CDEs within TCGA clinical cancer study domains.
• The tool detected various CDE modeling errors, validating its effectiveness.
• Demonstrated the feasibility of using compositional expressions for automated QA of terminological annotations.

Conclusions:

• Compositional expressions facilitate the reuse of existing ontology codes for defining new domain concepts.
• The developed system provides an automated mechanism for quality assurance of CDE terminological annotations.
• This approach significantly improves the reliability and consistency of cancer research data.