Deidentification of free-text medical records using pre-trained bidirectional transformers

Abstract

A neural network model for removing protected health information from clinical notes.

Publication
Proceedings of the ACM Conference on Health, Inference, and Learning