Annotating and recognising named entities in clinical notes

Abstract

This paper presents ongoing research in clinical information extraction. This work introduces a new genre of text which are not well-written, noise prone, ungrammatical and with much cryptic content. A corpus of clinical progress notes drawn form an Intensive Care Service has been manually annotated with more than 15000 clinical named entities in 11 entity types. This paper reports on the challenges involved in creating the annotation schema, and recognising and annotating clinical named entities. The information extraction task has initially used two approaches: a rule based system and a machine learning system using Conditional Random Fields (CRF). Different features are investigated to assess the interaction of feature sets and the supervised learning approaches to establish the combination best suited to this data set. The rule based and CRF systems achieved an F-score of 64.12% and 81.48% respectively. © 2009 ACL and AFNLP.