Concept-based AI interpretability in physiological time-series data: Example of abnormality detection in electroencephalography
TCAV has the potential to improve interpretability of deep learning applied to multi-channel signals as well as to detect possible biases in the data. Still, further work on developing the strategies for concept definition and validation on clinical physiological time series is needed to better understand how to extract clinically relevant information from the concept sensitivity scores.
Authors
- Henner Koch (ORCID: https://orcid.org/0000-0002-6883-3071)
- Alexander Brenner
- Rainer Röhrig (ORCID: https://orcid.org/0000-0002-0032-5118)
- Julian Varghese (ORCID: https://orcid.org/0000-0002-7206-3719)
- Felix Knispel
- Ekaterina Kutafina (ORCID: https://orcid.org/0000-0002-3430-5123)
- Florian P. Fischer (ORCID: https://orcid.org/0009-0006-4236-8403)
- Peter Rossmanith (ORCID: https://orcid.org/0000-0003-0177-8028)
- Yvonne Weber (ORCID: https://orcid.org/0000-0002-0806-5592)
Institutions
- University of Cologne (DE)
- University of Münster (DE)
- Universitätsklinikum Aachen (DE)
- University Hospital Cologne (DE)
- RWTH Aachen University (DE)
Publication Details
- Journal
- Computer Methods and Programs in Biomedicine
- Published
- 2024-09-30
- DOI
- https://doi.org/10.1016/j.cmpb.2024.108448
- Citations
- 5
- Primary Topic
- EEG and Brain-Computer Interfaces
- Type
- article
- Field-Weighted Citation Impact
- 1.71
Funders
- Westfälische Wilhelms-Universität Münster