BQS Registry Search: Identifying Medical Registries as Potential Data Sources

Author: Eveline Prochaska, Dresden University of Technology, 16 July 2026

Health data provide an essential foundation for the development and evaluation of data-driven methods in medicine. This is particularly relevant for research projects that aim to develop, test, or transfer artificial intelligence methods into medical application contexts.

Before data can be analysed, researchers must first determine which data sources may be suitable for a specific research question. In addition to clinical information systems, research data centres, cohorts, and claims data, medical registries can play an important role. They collect structured information on specific diseases, treatments, medical devices, or areas of healthcare and can therefore provide valuable foundations for medical research.

Within the KIMed – Network for Artificial Intelligence in Medicine project, we are addressing, among other topics, the question of how information about existing health data can be made easier to find, understand, and use for research projects. One publicly accessible resource that can support initial orientation is the registry search provided by the BQS Institute for Quality and Patient Safety [1].

The BQS Registry Search is a publicly accessible information resource on medical registries in Germany [2]. Its search interface allows users to search registered data collections by disease, medical specialty, or other terms.

The results provide structured descriptions of the registries included in the database. These may contain information such as:

• name and thematic focus of the registry,
• underlying diseases or indications,
• purpose and objectives of the registry,
• responsible organisations and registry operators,
• duration or current status,
• participating or reporting institutions,
• approximate number of recorded cases.

The platform therefore does not provide health data in the strict sense. Instead, it provides metadata, meaning descriptive information about existing registries and their contents.

When planning a research project, the first step is to determine whether suitable data are available and under which conditions they may be used.

For AI-related research, relevant questions include:

• Which patient groups and diseases are covered?
• Which clinical characteristics, examinations, or treatment outcomes are documented?
• How large and complete is the data set?
• Over what period were the data collected?
• Were the data recorded once or longitudinally?
• Do the data originate from multiple institutions?
• Under which conditions can the data be accessed for research purposes?
• Can the data be linked with other data sources?

The BQS Registry Search does not answer these questions in full. However, it can provide an initial structured overview of the German registry landscape. Researchers can use it to identify registries that are relevant to a project idea and then contact the responsible organisations directly.

The platform is therefore particularly useful during the early stages of a research project: for exploring possible data sources, refining a research question, and preparing more detailed data enquiries.

An entry in the Registry Search does not automatically mean that the underlying data are freely available or can be used directly in a research project. Medical registries are subject to organisational, legal, ethical, and data protection requirements.

After a potentially suitable registry has been identified, further steps are generally required. These typically include:

1. reviewing the available registry description,
2. contacting the registry operator,
3. clarifying the available data types and variables,
4. assessing data quality and completeness,
5. coordinating the scientific use concept,
6. clarifying data protection, consent, and ethical requirements,
7. where applicable, submitting a formal data access request.

The development of AI models involves additional requirements. Relevant considerations include the representativeness of the data, potential biases, documentation of data collection, handling of missing values, and whether the data are suitable for training, validation, or external evaluation.

The Registry Search should therefore not be understood as a data access portal. Rather, it is a starting point for structured research.

The discoverability of suitable health data is a central challenge in medical informatics [3]. Relevant data sets often exist, but they are distributed across different institutions and described according to varying professional, technical, and organisational criteria.

To make health data usable for research and innovation, not only the data themselves but also information about those data must be available [4]. Such metadata may describe:

• which data were collected,
• the healthcare context in which they were generated,
• the population represented,
• the standards and terminologies used,
• the temporal and geographic coverage,
• any restrictions on use,
• the responsible contact for enquiries.

Metadata allow researchers to assess the potential suitability of a data source before initiating complex application and access procedures [4]. They are therefore an important prerequisite for transparent, reproducible, and efficient data-driven research.

The KIMed – Network for Artificial Intelligence in Medicine project aims to further develop infrastructures and application opportunities for medical AI in Saxony. A central aspect is how existing health data and data-related expertise can be better identified and made accessible for research, development, and transfer.

This involves more than the technical development of data and analytics infrastructures. It is equally important to provide understandable information about existing data sources and to support researchers, healthcare institutions, and other project partners in navigating the data landscape.

Presenting publicly accessible research tools such as the BQS Registry Search is therefore part of a broader consideration of health data ecosystems. It illustrates how information about existing data resources can be consolidated and made discoverable.

For AI projects, this can be a first step towards developing realistic, data-informed research proposals. However, the actual usability of a registry must always be assessed in consultation with the responsible registry operators and in consideration of the relevant legal and methodological requirements.

The BQS Registry Search provides an overview but does not replace a detailed assessment of a registry. The value of each entry depends on how current and complete the information provided is.

Furthermore, the existence of a registry alone does not allow reliable conclusions about its specific data quality or scientific usability. The following aspects are not necessarily apparent from the registry description:

• completeness and currency of the data sets,
• scope of the variables collected,
• proportion of missing data,
• coding systems and terminologies used,
• representativeness of the included population,
• options for data linkage,
• specific access and conditions of use.

The development of AI methods also involves specific requirements that go beyond a general registry description. A large data set is not automatically suitable for machine learning. Relevant factors include the quality of the outcome variables, the comparability of data collection procedures, and sufficient documentation of the context in which the data were generated.

The BQS Registry Search is a useful orientation tool for researchers looking for medical registries and potential health data sources in Germany. It does not provide patient-level data or direct access to data. Its value lies in the structured description of existing registries.

For projects in medical informatics and artificial intelligence, it can support researchers in:

• obtaining an overview of the registry landscape for a research topic,
• identifying potentially relevant data sources,
• aligning research questions with available data resources,
• preparing targeted enquiries to registry operators.

The platform can therefore represent an initial step in a longer process—from searching for suitable health data and assessing their usability to conducting analyses under appropriate legal, methodological, and technical conditions.

Within the KIMed project, we aim to make these pathways to health data more transparent and to support exchange between data providers, researchers, and future users of medical AI.