Our work focuses on machine learning for biomedical data and human language, and on educating the next generation of AI experts at Aachen University of Applied Sciences.
Biomedical AI
Our AI model AnySleep analyzes sleep from flexible combinations of brain-wave and eye-movement signals. Its high-resolution predictions reveal brief changes obscured by conventional 30-second scoring. Published in npj Digital Medicine.
Explore AnySleepNatural language
We adapt language models to specialized German-language tasks with limited labeled data, from identifying toxic comments to detecting supportive language. Our approaches achieved first place at GermEval 2021 and 2025.
Explore language modelingResearch in practice
SomnoBot lets researchers analyze sleep recordings in a web browser. Our AI models AnySleep and SUMO run locally on their computer, without uploading patient data or installing specialist software.
Explore SomnoBot🎉 We celebrate the publication of our AnySleep paper in npj Digital Medicine! Our open-source model analyzes sleep from flexible combinations of EEG and EOG signals and provides sleep-stage predictions at adjustable temporal resolutions. We trained and validated AnySleep on more than 20,000 overnight recordings across 28 datasets. 📄 Read the paper, explore the code, and try it out online.