De-identification of medical records using conditional random fields and long short-term memory networks

TL;DR

This paper compares CRF and LSTM models for de-identifying psychiatric records, demonstrating that LSTMs outperform CRFs with higher accuracy in identifying protected health information.

Contribution

It introduces a novel LSTM-based approach for de-identification and compares its performance with traditional CRF models on clinical text.

Findings

LSTM system achieved an i2b2 F1 score of 89.86%.

LSTMs outperformed CRFs in PHI detection accuracy.

Pre-processing improved model performance.

Abstract

The CEGS N-GRID 2016 Shared Task 1 in Clinical Natural Language Processing focuses on the de-identification of psychiatric evaluation records. This paper describes two participating systems of our team, based on conditional random fields (CRFs) and long short-term memory networks (LSTMs). A pre-processing module was introduced for sentence detection and tokenization before de-identification. For CRFs, manually extracted rich features were utilized to train the model. For LSTMs, a character-level bi-directional LSTM network was applied to represent tokens and classify tags for each token, following which a decoding layer was stacked to decode the most probable protected health information (PHI) terms. The LSTM-based system attained an i2b2 strict micro-F_1 measure of 89.86%, which was higher than that of the CRF-based system.