Landscape analysis using a health-related data catalogue matrix

This document presents the findings of the extended landscape analysis performed in HealthyCloud WP3. The aim of HealthyCloud WP3 is to carry out a landscape analysis of available health-related data infrastructures, in order to capture the European health data collections available for research purposes, evaluate their FAIRness levels and determine the feasibility to perform individual level data linkages.

To perform this landscape analysis, Task 3.1 focused on collecting information on available health-related data infrastructures, including their governance, health-related domains covered, structure of the data stored, quality assurance of the datasets and the adherence to the FAIR principles (Findability, Accessibility, Interoperability and Re-usability), among others.

Initially, to collect this information, a survey was designed in collaboration with the leaders of WP4, in the form of a catalogue matrix. The previous deliverable D3.1 'Landscape analysis of FAIRness levels of health-related data using a catalogue matrix' presented the initial landscape analysis performed with the catalogue matrix, which focused specifically on the scope of the HealthyCloud use cases on atrial fibrillation and cancer.

The deliverable D3.3 'Landscape analysis using a health related-data catalogue matrix' presents the subsequent extension of that landscape analysis. The extension was achieved through a collaboration between the WP3 team and the Population Health Information Research Infrastructure (PHIRI) project, which has developed the European Health Information Portal (HIP), a one stop shop for services for researchers, including a metadata catalogue of health data collections.

In terms of methodology, the HIP metadata template of data sources was compared to the catalogue matrix developed by HealthyCloud, allowing the identification of properties common to both as well as essential properties missing from the HIP metadata template, which were proposed to be added. Subsequently, the HIP metadata template was sent to HealthyCloud partners so that they could add a record of their data collection. The landscape analysis was performed by analysing all new and existing data source records in the HIP (over 330). The key properties to be analysed were identified to be in line with the previous methodology in D3.1. Finally, a FAIRness evaluation was carried out of the new records made by HealthyCloud partners, using the FAIRness evaluation tool developed in D3.1.

The results section of this deliverable D3.3 shows the analysis of the following properties: type of information, geographical coverage, target population, access information, updating periodicity, personal identifier, level of aggregation, linkage possibility, permanent identifier of the data source. Over 330 data collections were analysed. In addition, six additional properties were analysed for the data collections added by HealthyCloud partners: type of data, anonymisation, community standards used to structure data, data format for exchange, metadata record, unique identifier for metadata. These properties were added as they are important properties to determine the FAIRness and linkability of data. The results of the FAIRness evaluation of the new records using the HealthyCloud FAIRness assessment tool are also presented.

Overall, the landscape analysis demonstrates that health-related data collections in Europe are highly heterogeneous. Most of the data collections currently included in the HIP are national or regional, but the collaboration with HealthyCloud allowed the addition of European level data infrastructures. This collaboration benefited both projects, allowing the completion of the HealthyCloud landscape analysis covering over 330 data collections, but also allowing the enrichment and future improvement of the HIP's European health-related metadata catalogue.