DEDUCE: de-identification method for Dutch medical text


This page contains a demo of the DEDUCE method for de-identification of Dutch medical text (v1.0.2).

Currently, DEDUCE annotates and de-identifies information in the following Protected Health Information (PHI) categories:
  1. Person names, including initials
  2. Geographical locations
  3. Names of institutions
  4. Dates
  5. Ages
  6. Patient numbers
  7. Phone numbers
  8. URLs and e-mail addresses

A detailed and techinical description of the method and its validation scores can be found in Menger et al., 2017. Some options to hand tailor the method to your specific problem are also described there. The project source code and installation manual can be accessed on Github.


In the text box below, paste your own medical text that is written in Dutch, and in the other input fields optionally add information about the patient. This server is (definitely) not designed to be secure enough to handle any actual patient data - so make sure to only use dummy data for demoing!

First name(s)
Given name