Clinical Concept Extraction with Contextual Word Embedding

TL;DR

This paper presents a new clinical concept extraction model that leverages domain-specific contextual word embeddings and a bidirectional LSTM-CRF architecture, achieving state-of-the-art results on the I2B2 2010 dataset.

Contribution

The study introduces a novel combination of domain-specific contextual embeddings with a bidirectional LSTM-CRF for improved clinical concept extraction.

Findings

Achieved 3.4% higher F1-score than previous state-of-the-art models.

Outperformed baseline models on the I2B2 2010 dataset.

Demonstrated effectiveness of domain-specific contextual embeddings in clinical NLP.

Abstract

Automatic extraction of clinical concepts is an essential step for turning the unstructured data within a clinical note into structured and actionable information. In this work, we propose a clinical concept extraction model for automatic annotation of clinical problems, treatments, and tests in clinical notes utilizing domain-specific contextual word embedding. A contextual word embedding model is first trained on a corpus with a mixture of clinical reports and relevant Wikipedia pages in the clinical domain. Next, a bidirectional LSTM-CRF model is trained for clinical concept extraction using the contextual word embedding model. We tested our proposed model on the I2B2 2010 challenge dataset. Our proposed model achieved the best performance among reported baseline models and outperformed the state-of-the-art models by 3.4% in terms of F1-score.