A FAIR data model for PRISMA (Personalised RISk-based MAmmascreening) Study

In the Netherlands, women aged 50-75 years are invited to receive breast cancer screening every two years. The PRISMA (Personalised RISk-based MAmmascreening) study was designed to investigate the added value of risk-based mammography screening. 43,000 screened women completed a web-based questionnaire comprising established risk factors and PROMs for breast cancer. There is no universally accepted data model for the collection of breast cancer risk factors.

Objective:

• To reduce ambiguity of data schemas in the domain and increase secondary use of client-reported outcomes through FAIRification, i.e. ensuring data are Findable, Accessible, Interoperable, and Reusable.
•  To create a FAIR semantic data model out of the PRISMA questionnaire.

Solution:

Step 1. 43,000 screened women completed a web-based questionnaire comprising established risk factors and PROMs for breast cancer.

Step 2. After several inventory meetings with different stakeholders, a consensus was reached on which data elements were important criteria to discover, share and reuse the PRISMA data. The resulting 67 data elements were grouped into 15 main classes

Linked Data representations for each CDE were constructed, by mapping to existing ontological terms

Step 3. The data elements identified in the PRISMA study can be instantiated according to the core elements (role, entity, process, measurement, attribute) in SIO and connected using the established property.

Added Value:

• FAIR data model, a potential template for breast cancer research groups to other PROMs and Real-World Experience questionnaires.

Next steps:

• create use cases that the PRISMA semantic data schema can support
• consider the interoperability with other standard, like the electronic health record (EHR)