Organizing data quality assessment of shifting biomedical data

Area of Science:

• Biomedical informatics
• Data science
• Healthcare analytics

Background:

• Low biomedical data quality (DQ) negatively impacts clinical decisions and evidence-based research outcomes.
• Existing DQ assessment methods often overlook the dynamic nature of data and concept relationships.
• Lack of consensus on standardized DQ dimensions and their interactions hinders robust assessment.

Purpose of the Study:

• To propose a novel framework for biomedical data quality assessment that accounts for shifting data concepts.
• To identify key characteristics and requirements for future DQ research in the biomedical domain.
• To establish a unified set of data quality dimensions for comprehensive DQ evaluation.

Main Methods:

• Conceptualizing a framework for biomedical DQ assessment considering data dynamism.
• Defining a comprehensive set of DQ dimensions.
• Developing the Data Quality Vector as a foundational element for DQ algorithms.

Main Results:

• A proposed organization for biomedical DQ assessment is presented.
• The Data Quality Vector is defined, integrating nine key DQ dimensions: completeness, consistency, duplicity, correctness, timeliness, spatial stability, contextualization, predictive value, and reliability.
• This vector serves as a basis for developing advanced DQ assessment tools.

Conclusions:

• The Data Quality Vector provides a unified and comprehensive approach to biomedical data quality assessment.
• This framework addresses limitations in current DQ methods by incorporating data concept shifts.
• The defined DQ dimensions and vector are foundational for future development of DQ assessment algorithms and platforms.