Boundary identification of events in clinical named entity recognition

TL;DR

This paper explores how sequence labeling models, specifically conditional random fields, combined with post-processing techniques, can improve the accuracy of boundary detection in clinical named entity recognition, which is crucial for clinical decision support.

Contribution

It demonstrates that sequence labeling and post-processing significantly enhance boundary identification accuracy in clinical event recognition.

Findings

Sequence labeling improves boundary detection accuracy.

Post-processing techniques further enhance recognition precision.

CRFs outperform previous methods in boundary identification.

Abstract

The problem of named entity recognition in the medical/clinical domain has gained increasing attention do to its vital role in a wide range of clinical decision support applications. The identification of complete and correct term span is vital for further knowledge synthesis (e.g., coding/mapping concepts thesauruses and classification standards). This paper investigates boundary adjustment by sequence labeling representations models and post-processing techniques in the problem of clinical named entity recognition (recognition of clinical events). Using current state-of-the-art sequence labeling algorithm (conditional random fields), we show experimentally that sequence labeling representation and post-processing can be significantly helpful in strict boundary identification of clinical events.