Visit to UC Berkeley
Visit to the University of California Berkeley
Research in Siegen Goes International: .MIGS Visits UC Berkeley
On May 15, 2026, research associates from the research group for Medical Informatics and Graph-based Systems (.MIGS) at the University of Siegen visited the University of California, Berkeley. During the visit with Professor Alam (Professor of Mechanical Engineering), current research projects were presented and discussed.
The exchange centered on two projects at the intersection of artificial intelligence, medicine, and data-driven decision support, sharing the common goal of improving medical processes, supporting reliable diagnostics, and addressing data privacy requirements along the way.
The visit opened with the "FACE" project. Cardiac arrhythmias and other abnormalities in long-term ECGs can be critical to patient safety, yet manual evaluation is time-consuming and prone to error: 24 hours of recording correspond to more than 100,000 heartbeats that must be assessed, with an average error rate of around 25 percent. FACE addresses this challenge with artificial intelligence: the project plans an automated ECG evaluation system in which AI models are trained to detect noise and classify arrhythmias. A particular focus lies on a privacy-friendly edge architecture, in which sensitive patient data is processed locally. In collaboration with medical practices, clinics, and device manufacturers, the integration of this system into everyday medical practice will be tested.
Two PhD research topics were presented next. The first focuses on predicting disease progression based on medical reports and patient records using large language models combined with knowledge graphs. The goal is to make medical information contained in unstructured text data usable in order to better estimate the course of a disease.
The second doctoral topic deals with the analysis of continuous vital sign data using artificial intelligence methods: both on the edge and in the cloud. The aim is to reliably detect patterns and changes in long-term measurement data, further strengthening data-based support in medicine going forward.
The exchange at UC Berkeley provided valuable input and feedback for further research. Overall, the projects presented show how AI in medicine can be designed to be not only powerful but also robust, quality-assured, and privacy-compliant, from cardiac diagnostics (FACE) to predictive approaches and intelligent analysis of continuous health data.
