DeID/ De-identification of clinical narrative data in French, German and Italian (DeID)

Project consortium: Prof. Christian Lovis (university hospitals of Geneva and university of Geneva), Dr. Vasiliki Foufi (university hospitals of Geneva and university of Geneva), Dr. Fabio Rinaldi (university of Zürich).

Main achievements

This project aims at providing a method for document de-identification based on opposable technologies, meaning that the deidentification processes and results could be:

  • Interpretable
  • Explainable
  • Correctable

In order to improve respect of the regulatory framework, including accountability and liability.

Reusable infrastructure and datasets

Case analysis and pilot implementation

Analysis of existing methodologies and their properties has been achieved and concluding that rule-based approaches was overall fulfilling best the requirements.

There are no existing rules formally defining which data is considered as being «identifying» data in Switzerland. However, it has been found that such rules exist in the USA, under the US Health Portability and Accountability Act (HIPAA). The HIPAA defines a legal framework and a technical guidance for using and disclosing health data and a list of elements that should be blurred to achieve de-identification of medical documents.

Available resources


The elements around understanding of de-identification in medicine, the HIPAA ruling and adaptation in Switzerland as well as their pilot implementation using finite state automata has been widely shared during several meetings and is available on demand (Contact: Prof. Christian Lovis)

Follow up projects – continuation – next steps

The project has ended at the end of the SPHN funding phase as national contribution. It has been continued at HUG with HUG funding to implement a full de-identification pipeline for documents.

Watch the SPHN webinar

The DeID infrastructure project is a small project that is intended to clarify the problem of document narratives deidentification in medicine. This field has been the subject of extensive work and publications for the last 20 years, with all types of approaches: knowledge-driven, rule-based, statistical and probabilistic.

References on Pubmed

  • Rochat J, Gaudet-Blavignac C, Del Zotto M, Ruiz Garretas V, Foufi V, Issom D, Samer C, Hurst S, Lovis C. Citizens' Participation in Health and Scientific Research in Switzerland. Stud Health Technol Inform. 2020 Jun 16;270:1098-1102.doi: 10.3233/SHTI200332. PMID: 32570551.
  • Foufi V, Gaudet-Blavignac C, Chevrier R, Lovis C. De-Identification of Medical Narrative Data. Stud Health Technol Inform. 2017;244:23-27. PMID: 29039370.
  • Chevrier R, Foufi V, Gaudet-Blavignac C, Robert A, Lovis C. Use and Understanding of Anonymization and De-Identification in the Biomedical Literature: Scoping Review. J Med Internet Res. 2019 May 31;21(5):e13484. doi: 10.2196/13484. PMID: 31152528; PMCID: PMC6658290.

Disclaimer: The contents on this website are intended as a general source of information and have been provided by the project PIs. The SPHN Management Office is not responsible for its accuracy, validity, or completeness.

Scroll to Top