A Transformer-Based Pipeline for German Clinical Document De-Identification
We trained and evaluated transformer models to detect sensitive information in German real-world pathology reports and progress notes. By defining an annotation scheme tailored to the documents of the investigating hospital and creating annotation guidelines for staff training, a further experimental study was conducted to compare the models with humans. These results showed that the best-performing model achieved better overall results than two experienced annotators who manually labeled 100 clinical documents.
Authors
- René Hosch (ORCID: https://orcid.org/0000-0003-1760-2342)
- Christoph M. Friedrich (ORCID: https://orcid.org/0000-0001-7906-0038)
- Giulia Baldini (ORCID: https://orcid.org/0000-0002-5929-0271)
- Ahmad Idrissi-Yaghir (ORCID: https://orcid.org/0000-0003-1507-9690)
- Felix Nensa (ORCID: https://orcid.org/0000-0002-5811-7100)
- Kamyar Arzideh (ORCID: https://orcid.org/0009-0005-6074-804X)
- Philipp Winnekens
Institutions
- Dortmund University of Applied Sciences and Arts (DE)
- Institut für Medizinische Informatik, Biometrie und Epidemiologie (DE)
- Essen University Hospital (DE)
Publication Details
- Journal
- Applied Clinical Informatics
- Published
- 2025-01-01
- DOI
- https://doi.org/10.1055/a-2424-1989
- Citations
- 4
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- article
- Field-Weighted Citation Impact
- 2.62
Funders
- Deutsches Zentrum für Luft- und Raumfahrt