CDEGenerator: an online platform to learn from existing data models to build model registries

Area of Science:

• Medical Informatics
• Data Science
• Clinical Research Informatics

Background:

• Harmonizing medical variables in clinical and epidemiological studies is crucial for data consistency.
• Existing data sets are challenging to search and reuse for specific study needs.
• A gap exists in efficiently comparing and customizing published data models for new research databases.

Purpose of the Study:

• To implement an Internet-based method for rapid comparison of published medical data models.
• To enable reuse, customization, and harmonization of item catalogs during research database development.
• To streamline the early planning phase of research databases through data model comparison.

Main Methods:

• Established a European information infrastructure with a collection of medical data models.
• Developed the CDEGenerator analysis module for systematic data model comparison.
• Assessed usability with eight external medical documentation experts using the System Usability Scale.

Main Results:

• The CDEGenerator module offers multilingual comparisons of complex clinical trial eligibility criteria.
• The tool demonstrated "good usability" with a mean System Usability Scale score of 75.0.
• User-tailored models are exportable to formats like XLS, REDCap, and the Operational Data Model.

Conclusions:

• The online tool provides user-friendly methods for reusing and comparing standardized data items.
• It facilitates learning from existing models to design harmonized research databases.
• Enables the creation of a blueprint for effective research database development.