metadata
language:
- en
pipeline_tag: token-classification
tags:
- medical
Protected health information (PHI) anonymization tool. Fine-tuned on the i2b2 2014 training dataset from the pretrained roberta-base model.
Anonymizes according to the i2b2 2014 standard, including all ages, locations and organizations, dates (including lone years), names, professions, identification numbers, and contact information.
Model released with the approval of Informatics for Integrating Biology & the Bedside.