The ETHOS-AI research group focuses on developing efficient, trustworthy, and robust Artificial Intelligence (AI) methods for high-impact applications, with a particular emphasis on healthcare and biomedical systems. The group addresses key challenges in modern AI, including reliability, interpretability, uncertainty representation, human-centred design, and safe integration of AI technologies into clinical practice.
A central research theme is developing uncertainty-aware AI approaches that enable models not only to generate predictions but also to quantify their confidence, identify unreliable outputs, and support appropriate human expert involvement. The group investigates hierarchical and patient-level uncertainty modelling, explainable AI, robust machine learning, and selective prediction strategies to improve the safety and transparency of AI-driven healthcare solutions.
ETHOS-AI combines AI development with User Experience (UX) design and human–AI interaction research for medical applications. The group studies how clinicians and healthcare professionals interact with AI systems, how AI-generated information should be presented, and how interfaces can support trust, understanding, and effective clinical decision-making. This human-centred approach ensures that AI technologies are not only technically accurate but also usable, acceptable, and aligned with healthcare workflows.
Research applications include medical image analysis, ophthalmology (glaucoma, keratoconus, retinal imaging), neurodegenerative disease progression modelling, and AI-supported clinical decision systems. The group brings together expertise from artificial intelligence, machine learning, uncertainty quantification, medical imaging, UX research, and clinical domains through collaborations with international academic and healthcare partners.
The group brings together researchers from FIIT STU, the University of Chester, the University of Southampton, MIT, the University of Liverpool, the University of Brescia, the University of Lancaster, and clinical partners. ETHOS-AI is supported through competitive research funding, including APVV, MISTI MIT, Siemens, and European investment programmes.
Trustworthy Artificial Intelligence, Uncertainty Quantification, Explainable AI, Robust Machine Learning, Medical Artificial Intelligence, Human–AI Interaction, UX Design for Healthcare, Clinical Decision Support, Medical Image Analysis, Personalised Healthcare AI
|
Gabriela Czanner Associate Professor e-mail: gabriela.czanner[at]stuba.sk Deputy head of the research group |
|
|
Silvester Czanner Associate Professor e-mail: silvester.czanner[at]stuba.sk |
| Her research interests include 1) Design of clinical studies and experiments for association studies as well as for development and testing of Artificial Intelligence, 2) Hierarchical statistical modelling for spatial and longitudinal data, including medical images, 3) Quantification of risk and uncertainty and how to use them for decision making. |
|
His research interests are: 1) designing efficient methods to learn from multimodal data streams, where information come as different modalities such as images, text, sensory data, etc.; 2) developing resource-efficient methods to reduce substantial costs of learning from massive datasets; and 3) designing reliable and safe learning algorithms with rigorous guarantees for safety-critical systems. | ||
|
Martina Bilichová Doctoral Student e-mail: martina.billichova[at]stuba.sk |
|
|
Michal Lüley Doctoral Student e-mail: michal.luley[at]stuba.sk |
|
|
|
|
||
|
Dmytro Furman Doctoral Student e-mail: dmytro.furman[at]stuba.sk |
|
|
|
|
|
|
|
||
