Published on April 08, 2025
Data sovereignty and efficiency: Kadi4Mat for the digital lab
Behind scientific and technical achievements there is often a phase of experimentation. In the laboratory, researchers, engineering specialists and analytical minds come together to work on samples and experimental setups - measuring, documenting and scrutinizing results - always in search of new insights. Each experiment generates a wealth of data: Measured values, reaction times, material properties, image-based test results and more. Only by analyzing these data sets can patterns, correlations and scientifically sound conclusions be drawn. This is why electronic laboratory notebooks (ELNs), a kind of logbook or test protocol, are standard in many laboratories. A team led by Dr. Michael Selzer from the Institute of Nanotechnology (INT) at KIT has developed the Kadi4Mat platform (Karlsruhe Data Infrastructure for Materials Science), which makes laboratory documentation and analysis faster, smarter and more sustainable. Kadi4Mat goes far beyond traditional electronic lab notebooks.
Creating knowledge from data
Not every experiment leads to the desired result, but every experiment provides valuable data. Research data management is therefore crucial in order to make data-based research results comprehensible, reproducible and usable in the long term. A virtual research environment such as Kadi4Mat supports those working in research and development in doing just that. “Researchers used to have to laboriously record their work manually, but later electronic lab notebooks made things easier,” recalls computer scientist Selzer, who used to work in materials simulation at KIT. His desire for even more efficiency in materials testing led to a new passion: "I changed my field of research - from materials science to research data science. It's exciting for me because it's a completely new field of research." As part of various joint projects and the NFDI4ING (National Research Data Infrastructure for Engineering Sciences), the scientific work has resulted in the now available platform Kadi4Mat, or Kadi for short. NFDI4ING is a consortium within the BMBF-funded NFDI initiative with the aim of establishing a sustainable infrastructure for research data.
FAIR instead of data frustration
"A classic ELN documents laboratory tasks. I like to call Kadi4Mat ELN 2.0: it's about documenting, controlling and evaluating the research process as a whole - right through to automation. We focus on workflows. To do this, we need to understand exactly which steps are necessary to implement good research data management in the various disciplines. Which devices do I need for an experiment? Which steps are necessary and in which order? And how do I use the devices? The entire process generates data that is relevant and ideally recorded automatically. We align the infrastructure to this," says Selzer, describing the complexity. It is particularly important to capture data simply and as automatically as possible and to store it in a structured way: the more data there is, the more important it is to catalog and classify the data accordingly so that you don't have to search for the proverbial needle in the haystack later on. In order to store information sensibly, make it available in the long term and relate it to each other, Kadi not only provides a repository, but also an easy-to-use tool: the software searches, analyses, visualizes and compares the accumulated data efficiently in order to maximize the information gained from scientific work in accordance with the established FAIR principles. "A classic ELN documents laboratory tasks. I like to refer to Kadi4Mat as ELN 2.0: it's about documenting the research process as a whole. "These state that research data should be findable, accessible, interoperable and reusable. Independently of Kadi, we are campaigning for a standard for data and exchange formats as part of the NFDI initiatives," reports Selzer.
The philosophy: modular and open
"With Kadi4Mat, we are creating an infrastructure that can document and automate the entire research process. It combines a next-generation ELN with powerful interfaces that ensure maximum flexibility and interoperability,“ summarizes developer Selzer and emphasizes: ”We didn't want to create another isolated solution, but rather offer an open and expandable platform that can be used in university and industrial research." Originally designed for materials science, the open source platform has developed into a modular system that can be individually adapted to the needs of users. In addition to the core component KadiWeb, there are various specialized modules, such as KadiStudio for workflow management or KadiAI for machine learning, which address different needs and can be combined. The basis is the KadiFS file system, which is connected to the laboratory devices and can therefore seamlessly store data without the need for further software adjustments. "You don't have to use everything, but you can. With open interfaces and modular features, Kadi4Mat can be used in any number of ways - from simple documentation to complex machine learning processes. Our aim is to create a platform that not only offers solutions for today's applications, but is also future-proof,“ emphasizes Selzer, one of the leading minds behind Kadi4Mat. ”We attach great importance to the software documentation. Users should be able to meet individual requirements via the programming interfaces developed. This enables diverse and flexible application options as well as a high degree of automation," reports Selzer from the developer's perspective.
A head start with Kadi4Mat
The combination of openness, flexibility and increased efficiency makes Kadi an unrivaled solution for research processes and collaboration between science and industry. “The biggest hurdle is often convincing the users,” Selzer admits openly and explains: "But once you use Kadi, you are enthusiastic and become a multiplier. The initial extra effort required for implementation pays off in day-to-day work. The feedback is correspondingly positive." Kadi4Mat offers numerous functions and scope for individualization. Application examples clearly show the added value: "In the POLIS (Post Lithium Storage) Cluster of Excellence, we support battery research with good research data management. With Kadi, test parameters can be fed directly into the connected devices and the tests can be started automatically while the researchers are already working on new issues. In the field of materials research, we were able to use KadiStudio to accelerate and improve the video evaluation of a micropillar test, i.e. the characterization of a material under mechanical stress. Whereas the manual evaluation of the video data previously took two hours, with Kadi it can be done at the touch of a button and within a few minutes. Thanks to the integration of AI, we can recognize patterns in large data sets more quickly and make the analyses more precise. This is not only a significant increase in efficiency, but also frees up time for more intensive research," says Selzer. With now several hundred users worldwide, the team at KIT is planning the next steps. The focus is on expanding the ecosystem and optimizing user-friendliness. "Our goal is to build an international community that focuses not only on using the platform, but also on developing it further. The first community workshop in Ulm in March 2025, which we organized together with DLR e.V., has already successfully set the course for this," says Selzer's update.
Data-driven innovation
With open interfaces and flexible workflows, the KIT platform ensures that research not only becomes more efficient, but also more sustainable and innovative. For companies, Kadi offers an ideal basis for an improved data infrastructure and optimized processes. “Our platform shows how application-oriented research and innovative technologies can successfully come together,” summarizes Selzer. By using a holistic system, research data from various sources can be collected in a standardized way and used for future projects. This supports the development of new products, materials, technologies or processes and improves collaboration between interdisciplinary teams.
Data security including
Despite the openness for various fields of application, data protection remains a key issue. In the scientific context, the sharing of data is desired and required, increasingly through open science. This contrasts with the prevailing industrial requirement to protect internal business information. “Data security is a critical factor, especially in an industrial environment,” emphasizes Selzer. Both perspectives are taken into account when implementing Kadi4Mat. In addition to a cloud version, the platform can be operated locally so that companies retain full control over their data. The team is also working on improving collaboration between decentralized instances, for example through an automated request function for data access and rights and role management. This allows collaboration at eye level, whether in basic research or in applied industry.
comments about this article
No comments