Patrik Thomas
Michalski.

Since July 2026, I have worked on computing and data analysis in the project, focusing on forecasts of regional care demand. I will add methods and results here when they are ready for publication.

The state of Schleswig-Holstein supports Pflege-Prognose+ with approximately €400,000 through its AI funding programme. Together with Pflege-Monitor+, the project is intended to make data from municipalities, care providers and state authorities usable for care planning.

I examined how an existing joint schema represents empirical research in machine learning and databases/data-intensive systems. Based on a follow-up survey, we extended the schema from three to five computer-science disciplines and adapted it to the additional fields.

The resulting paper for the fourth NFDIxCS Symposium at the INFORMATIK Festival 2026 has been published. My involvement from January to June 2026 was part of my work at the University of Bayreuth.

NFDIxCS supports research data management in computer science according to the FAIR principles: data and software should be findable, accessible, interoperable and reusable. One consortium concept is the Research Data Management Container, which bundles related research artefacts.

From October to December 2025, I applied data-science methods to data from the participating fields at the Datencampus and exchanged approaches with researchers in electrical engineering and physics.

In 2022, the state of Schleswig-Holstein awarded approximately €2 million to establish the Datencampus, initially for three years. Four tandems were intended to connect computer-science methods with research questions from other disciplines.

I developed a generator for synthetic force–displacement curves with controllable material, geometry and loading parameters. The associated publication examines the use of these data to train and evaluate sequence models.

For the subsequent study of a surrogate model for tool optimisation, I developed the multilayer-perceptron workflow, an explainability analysis and the optimisation loop. Proposed punch surfaces were reconstructed from the model parameters and checked in additional finite-element simulations. The manuscript is under review.

DFG Priority Programme 2422 investigates data-driven models of metal-forming processes. It combines process data with expert knowledge and simulations. Prof. Mathias Liewald at the University of Stuttgart coordinates the programme.

From January 2023 to March 2025, I worked on route planning and spatial data management. I investigated how the movement of other vessels can be taken into account when assessing collision risk and choosing routes.

Marispace-X was funded through Germany’s Gaia-X competition. Kiel University participated under the coordination of Prof. Dr. Matthias Renz. The consortium’s announced total budget was approximately €15 million.

The planned maritime data space connected edge, fog and cloud processing. Use cases included mapping seagrass meadows and legacy munitions, as well as using maritime data for shipping.