Joint inference for end-to-end coreference resolution for clinical notes

Prateek Jindal, Dan Roth, Carl Gunter

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Recent US government initiatives have led to wide adoption of Electronic Health Records (EHRs). More and more health care institutions are storing patients' data in an electronic format. These EHRs contain valuable information which can be used in important applications like Clinical Decision Support (CDS). So, Information Extraction (IE) from EHRs is a very promising research area. This paper presents a robust method for end-to-end coreference resolution for clinical narratives. For our experiments, we used the datasets provided by i2b2/VA team as part of i2b2/VA 2011 shared task on coreference resolution. One part of this data was annotated according to ODIE guidelines and another part was annotated according to i2b2 guidelines. We designed a global inference strategy for end-to-end coreference resolution which jointly determines the mention types and coreference relations between them. This technique avoids the problem of error-propagation which is common in pipeline systems. For pronominal resolution, we developed different strategies for resolving different pronouns. We report the best results to date on both ODIE and i2b2 data. We got the best results for both types of cases: (1) where gold mentions are already given and (2) for end-to-end coreference resolution. ODIE and i2b2 data are annotated quite differently. Best results on both types of data proves the robustness of our algorithm.

Original languageEnglish (US)
Title of host publicationACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics
PublisherAssociation for Computing Machinery, Inc
Pages192-201
Number of pages10
ISBN (Electronic)9781450328944
DOIs
StatePublished - Sep 20 2014
Event5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics, ACM BCB 2014 - Newport Beach, United States
Duration: Sep 20 2014Sep 23 2014

Publication series

NameACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics

Other

Other5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics, ACM BCB 2014
CountryUnited States
CityNewport Beach
Period9/20/149/23/14

Fingerprint

Electronic Health Records
Joints
Health
Clinical Decision Support Systems
Guidelines
Information Storage and Retrieval
Health care
Gold
Pipelines
Delivery of Health Care
Research
Experiments

Keywords

  • Coreference resolution
  • Health informatics
  • Integer programming
  • Joint inference
  • Natural language processing

ASJC Scopus subject areas

  • Health Informatics
  • Computer Science Applications
  • Software
  • Biomedical Engineering

Cite this

Jindal, P., Roth, D., & Gunter, C. (2014). Joint inference for end-to-end coreference resolution for clinical notes. In ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics (pp. 192-201). (ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics). Association for Computing Machinery, Inc. https://doi.org/10.1145/2649387.2649437

Joint inference for end-to-end coreference resolution for clinical notes. / Jindal, Prateek; Roth, Dan; Gunter, Carl.

ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics. Association for Computing Machinery, Inc, 2014. p. 192-201 (ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Jindal, P, Roth, D & Gunter, C 2014, Joint inference for end-to-end coreference resolution for clinical notes. in ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics. ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics, Association for Computing Machinery, Inc, pp. 192-201, 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics, ACM BCB 2014, Newport Beach, United States, 9/20/14. https://doi.org/10.1145/2649387.2649437
Jindal P, Roth D, Gunter C. Joint inference for end-to-end coreference resolution for clinical notes. In ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics. Association for Computing Machinery, Inc. 2014. p. 192-201. (ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics). https://doi.org/10.1145/2649387.2649437
Jindal, Prateek ; Roth, Dan ; Gunter, Carl. / Joint inference for end-to-end coreference resolution for clinical notes. ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics. Association for Computing Machinery, Inc, 2014. pp. 192-201 (ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics).
@inproceedings{c0950749e0e74926b827ef7411c3569d,
title = "Joint inference for end-to-end coreference resolution for clinical notes",
abstract = "Recent US government initiatives have led to wide adoption of Electronic Health Records (EHRs). More and more health care institutions are storing patients' data in an electronic format. These EHRs contain valuable information which can be used in important applications like Clinical Decision Support (CDS). So, Information Extraction (IE) from EHRs is a very promising research area. This paper presents a robust method for end-to-end coreference resolution for clinical narratives. For our experiments, we used the datasets provided by i2b2/VA team as part of i2b2/VA 2011 shared task on coreference resolution. One part of this data was annotated according to ODIE guidelines and another part was annotated according to i2b2 guidelines. We designed a global inference strategy for end-to-end coreference resolution which jointly determines the mention types and coreference relations between them. This technique avoids the problem of error-propagation which is common in pipeline systems. For pronominal resolution, we developed different strategies for resolving different pronouns. We report the best results to date on both ODIE and i2b2 data. We got the best results for both types of cases: (1) where gold mentions are already given and (2) for end-to-end coreference resolution. ODIE and i2b2 data are annotated quite differently. Best results on both types of data proves the robustness of our algorithm.",
keywords = "Coreference resolution, Health informatics, Integer programming, Joint inference, Natural language processing",
author = "Prateek Jindal and Dan Roth and Carl Gunter",
year = "2014",
month = "9",
day = "20",
doi = "10.1145/2649387.2649437",
language = "English (US)",
series = "ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics",
publisher = "Association for Computing Machinery, Inc",
pages = "192--201",
booktitle = "ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics",

}

TY - GEN

T1 - Joint inference for end-to-end coreference resolution for clinical notes

AU - Jindal, Prateek

AU - Roth, Dan

AU - Gunter, Carl

PY - 2014/9/20

Y1 - 2014/9/20

N2 - Recent US government initiatives have led to wide adoption of Electronic Health Records (EHRs). More and more health care institutions are storing patients' data in an electronic format. These EHRs contain valuable information which can be used in important applications like Clinical Decision Support (CDS). So, Information Extraction (IE) from EHRs is a very promising research area. This paper presents a robust method for end-to-end coreference resolution for clinical narratives. For our experiments, we used the datasets provided by i2b2/VA team as part of i2b2/VA 2011 shared task on coreference resolution. One part of this data was annotated according to ODIE guidelines and another part was annotated according to i2b2 guidelines. We designed a global inference strategy for end-to-end coreference resolution which jointly determines the mention types and coreference relations between them. This technique avoids the problem of error-propagation which is common in pipeline systems. For pronominal resolution, we developed different strategies for resolving different pronouns. We report the best results to date on both ODIE and i2b2 data. We got the best results for both types of cases: (1) where gold mentions are already given and (2) for end-to-end coreference resolution. ODIE and i2b2 data are annotated quite differently. Best results on both types of data proves the robustness of our algorithm.

AB - Recent US government initiatives have led to wide adoption of Electronic Health Records (EHRs). More and more health care institutions are storing patients' data in an electronic format. These EHRs contain valuable information which can be used in important applications like Clinical Decision Support (CDS). So, Information Extraction (IE) from EHRs is a very promising research area. This paper presents a robust method for end-to-end coreference resolution for clinical narratives. For our experiments, we used the datasets provided by i2b2/VA team as part of i2b2/VA 2011 shared task on coreference resolution. One part of this data was annotated according to ODIE guidelines and another part was annotated according to i2b2 guidelines. We designed a global inference strategy for end-to-end coreference resolution which jointly determines the mention types and coreference relations between them. This technique avoids the problem of error-propagation which is common in pipeline systems. For pronominal resolution, we developed different strategies for resolving different pronouns. We report the best results to date on both ODIE and i2b2 data. We got the best results for both types of cases: (1) where gold mentions are already given and (2) for end-to-end coreference resolution. ODIE and i2b2 data are annotated quite differently. Best results on both types of data proves the robustness of our algorithm.

KW - Coreference resolution

KW - Health informatics

KW - Integer programming

KW - Joint inference

KW - Natural language processing

UR - http://www.scopus.com/inward/record.url?scp=84920717426&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84920717426&partnerID=8YFLogxK

U2 - 10.1145/2649387.2649437

DO - 10.1145/2649387.2649437

M3 - Conference contribution

AN - SCOPUS:84920717426

T3 - ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics

SP - 192

EP - 201

BT - ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics

PB - Association for Computing Machinery, Inc

ER -