German Portal for Medical Research Data (FDPG)

Data Infrastructure for Research, Healthcare, and AI Development

Author: Dr Eveline Prochaska, Dresden University of Technology, 22 May 2026

Data-driven medicine depends fundamentally on the availability of high-quality, interoperable, and legally compliant data. In Germany, the German Portal for Medical Research Data (Forschungsdatenportal Gesundheit, FDPG) was established as a central application and coordination platform within a federated data infrastructure, providing a structured interface between clinical care and scientific data use [1].

From a medical informatics perspective, the FDPG serves as a governance and integration mechanism for the secondary use of routine clinical data, supporting research, industry, and education alike.

The FDPG is an outcome of the German Medical Informatics Initiative (MII), which has been funded by the Federal Ministry of Research, Technology and Space (BMFTR) since 2018 [2,3]. The initiative aims to make clinical routine data from German university hospitals available for research while systematically embedding interoperability, data protection, and ethical standards.

As part of the MII, Data Integration Centers have been established at participating university hospitals. These centers harmonize routine clinical data using a common core dataset and international interoperability standards such as HL7 FHIR, together with established terminologies and coding systems including ICD, LOINC, and, where applicable, SNOMED CT. The conceptual foundations of this federated architecture have been described by Semler et al. [4] and Haarbrandt et al. [5].

Within this framework, the FDPG acts as the central application and coordination platform for multicenter data use projects, while the data themselves remain decentralized within the respective Data Integration Centers [1].

The German Portal for Medical Research Data is not a centralized data repository and does not store clinical routine data itself. Instead, the data remain at the participating university hospitals within their respective Data Integration Centers. The portal functions as a coordinating application and governance platform [1].

Researchers submit a project proposal via the FDPG describing the research question, the required dataset, and the planned data protection and ethics measures. The application is coordinated across participating sites and reviewed by local Use-and-Access Committees, data protection offices, and, where required, institutional ethics committees [1,4].

Once approval has been granted, data extraction is performed locally within the respective Data Integration Centers. The data are generally pseudonymized and made available within secure analysis environments. Clinical routine data are not centrally pooled across all participating institutions.

Importantly, the data are generally not released as freely downloadable raw datasets. Instead, analyses are conducted within controlled secure environments or under clearly defined technical and organizational safeguards. This federated model of controlled data use reflects the principles of decentralized architecture and governance established within the German Medical Informatics Initiative [4].

This figure illustrates the multi-stage workflow from project proposal and application submission through cross-site review (Use-and-Access Committees, data protection review, and, where applicable, ethics approval) to controlled data extraction within the Data Integration Centers. It highlights the decentralized data architecture: clinical routine data remain at the participating university hospitals and are not centrally stored within the FDPG. Project-specific analyses are conducted in secure environments under defined technical and organizational safeguards.

For the scientific community, the German Portal for Medical Research Data enables multicenter analyses based on harmonized real-world data, i.e., clinical routine data generated during healthcare delivery. Such semantically standardized multicenter datasets are particularly valuable for health services research, epidemiological studies, and the validation of AI-based models [4,5]. In particular, multicenter AI development relies on semantically harmonized data structures, interoperable standards, and transparent governance processes.

At the same time, access to these data is intentionally regulated and subject to strict requirements. Data use is project-specific and purpose-limited, requiring a scientifically justified research question together with approved data protection and ethics concepts. Not every project qualifies for approval, particularly if it lacks clear scientific merit or an adequate legal basis.

Furthermore, several structural limitations should be considered:

• Data availability varies between participating sites.
• Data quality depends on local clinical documentation practices.
• Not all desired variables are included in the MII core dataset.
• Review and coordination procedures may result in time delays.

These restrictions also apply to industrial users. The FDPG is not an open data marketplace. Access for purely commercial purposes without a clear research context is generally not intended. Collaborations take place within clearly defined scientific and regulatory frameworks [1,4].

For AI development, this means that training and validation of models are feasible but require careful planning, formal application procedures, and close coordination with the participating institutions.

The German Portal for Medical Research Data exemplifies the structural transformation toward coordinated, data-driven healthcare. It demonstrates that access to clinical routine data is not merely a technical challenge but equally an organizational and regulatory one. Through its federated architecture, standardized processes, and transparent governance, the FDPG establishes a reliable framework for multicenter research. At the same time, its concepts of federated data use, interoperable standards, and secure analysis environments align closely with current European developments, including the European Health Data Space (EHDS) and future Secure Processing Environments.