A Concept Annotation System for Clinical Records

TL;DR

This paper introduces a novel system for annotating medical concepts in clinical records, combining multiple NER tools with a voting scheme, achieving high accuracy and offering accessible interfaces for integration.

Contribution

It presents the first publicly available clinical concept annotation system that integrates six NER tools with adjustable precision and recall capabilities.

Findings

Achieved an F-score of 82.1% in the i2b2 challenge

System is among top-ranking in concept annotation

Provides web and UIMA interfaces for easy use

Abstract

Unstructured information comprises a valuable source of data in clinical records. For text mining in clinical records, concept extraction is the first step in finding assertions and relationships. This study presents a system developed for the annotation of medical concepts, including medical problems, tests, and treatments, mentioned in clinical records. The system combines six publicly available named entity recognition system into one framework, and uses a simple voting scheme that allows to tune precision and recall of the system to specific needs. The system provides both a web service interface and a UIMA interface which can be easily used by other systems. The system was tested in the fourth i2b2 challenge and achieved an F-score of 82.1% for the concept exact match task, a score which is among the top-ranking systems. To our knowledge, this is the first publicly available clinical record concept annotation system.