A Suite of Natural Language Processing Tools ... - Semantic Scholar

19 downloads 7161 Views 141KB Size Report
A Suite of Natural Language Processing Tools Developed for the. I2B2 Project ... they were often domain-, institution- and application- specific. We have ... Finder. List of. Matching. Concepts. Section. Filter. Text. Tokenizer. Sentence. Splitter.
A Suite of Natural Language Processing Tools Developed for the I2B2 Project Sergey Goryachev, MS, Margarita Sordo, Ph.D., Qing T. Zeng, Ph.D. Decision Systems Group, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA ABSTRACT Textual medical records contain a wealth of information that needs to be extracted and / or indexed in order to be analyzed and interpreted by the automated tools. We have developed a collection of natural language processing (NLP) tools to extract various types of information from unstructured medical records. The generic NLP components, when assembled in pipelines and initialized with custom configuration parameters, become a powerful medical data mining instrument. We have successfully extracted such medical concepts as diagnoses, comorbidities, discharge medications, and smoking status from various types of medical records.

INTRODUCTION

Medical Report

Syntactic Parsing-based Method Section Splitter

Section Filter

A textual medical record is a rich source of clinical information. Although a number of NLP systems had demonstrated good accuracy in information extraction, they were often domain-, institution- and applicationspecific. We have developed a suite of NLP tools for the I2B2 (Informatics for Integrating Biology and the Bedside, a national center for biomedical computing) project to address a wide range of text processing needs. We took a modularized and parameterized approach in the software development and employed syntactic, statistical, template-based methods for different parsing tasks. This approach allows users to tailor the NLP tools to extract and index specific information from different domains and institutions.

METHODS AND RESULTS We have developed 11 modules for text report processing (Figure 1): 1. Section Splitter 2. Section Filter 3. Text Tokenizer 4. Part-of-Speech (POS) Tagger 5. Noun Phrase Finder 6. UMLS Concept Finder 7. Negation Finder 8. Regular Expression-based Concept Finder 9. Sentence Splitter 10. N-Gram Tool 11. Classifier (e.g. Smoking Status Classifier) These modules were applied to discharge summaries and outpatient notes from 2 institutions, Brigham and Women’s Hospital and Massachusetts General Hospital with minimum changes in the configuration files. They were also used to extract key data items from a set of medical error reports, which involved adding several new modules, but didn’t require any alteration of the original 11 modules.

Textual Medical Records

Text Tokenizer

Part-ofSpeech Tagger

Noun Phrases Finder

1

2

3

4

5

Template-based Method Regular Expressionbased Concept Finder

Collection of report sections

Discharge Medications

Statistical Method Sentence Splitter

Collection of tagged tokens

For example:

List of Matching Concepts

Collection of report sections matching some criteria

Collection of tokens (words)

8

9

10

Collection of sentences

N-Gram Tool

11

N-Gram frequencies

Classifier

For example:

UMLS Concept Finder

6

7

Collection of noun phrases

Classification

smoker, nonsmoker, past smoker, etc

Collection of UMLS concepts

Negation Finder

UMLS Concepts with Negation

For example:

Diagnoses or Comorbidities

Figure 1. NLP components for medical report processing assembled into pipelines for various information extraction tasks.

AMIA 2006 Symposium Proceedings Page - 931