Machine learning algorithms have vast potential in healthcare, but many advances in research fail to translate to clinical practice today. In part, this is due the the lack of representative medical data. But even with larger datasets, algorithms can fail to generalize because of biased datasets, or inadequate evaluation during development. Our research focuses of overcoming these challenges by studying methods for limited labelled data, such as transfer learning, and responsible practices for machine learning in health in general. Check out our publications.
Team
We are part of the DAta-intensive Systems and Applications (DASYA) group at ITU Copenhagen.
- Veronika Cheplygina – Associate professor
- Ralf Raumanns – PhD fellow (TU Eindhoven, 2019-now)
- Bethany Chamberlain – PhD fellow (2022 – now)
- Dovile Juodelyte – PhD fellow (2022 – now)
- Amelia Jiménez-Sánchez – Postdoc (2022 – now)
If you’d like to do research with us, please get in touch with Veronika.
Funded projects
MMC – Making Metadata Count – funded by Independent Research Council of Denmark, Inge Lehmann programm (2.9M DKK), 2022-2025. PI Veronika Cheplygina, with Amelia Jiménez-Sánchez
CATS – Choosing A Transfer Source for medical image classification – funded by Novo Nordisk Foundation Starting Package (3.5M DKK), 2022 – 2025. PI Veronika Cheplygina, with Bethany Chamberlain and Dovile Juodelyte
HINTS – From black box to intelligible machine learning for the accurate diagnosis of medical images – funded by Dutch Research Council (NWO). Principal investigator Ralf Raumanns, 2019-2024