Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A reliability study for evaluating information extraction from radiology reports.

G Hripcsak1, G J Kuperman, C Friedman

  • 1Columbia University, New York, New York, USA. hripcsak@columbia.edu

Journal of the American Medical Informatics Association : JAMIA
|March 27, 1999
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Urban air pollution and in vitro fertilization outcomes: A Canadian retrospective study.

Reproductive toxicology (Elmsford, N.Y.)·2026
Same author

MAO inhibitory activity of bromo-2-phenylbenzofurans: synthesis, <i>in vitro</i> study, and docking calculations.

MedChemComm·2018
Same author

Estimating summary statistics for electronic health record laboratory data for use in high-throughput phenotyping algorithms.

Journal of biomedical informatics·2018
Same author

Development of a battery of tests to measure attitudes and intended behaviours of dental students towards people with disability or those in marginalised groups.

European journal of dental education : official journal of the Association for Dental Education in Europe·2017
Same author

Patient-Centered Network of Learning Health Systems: Developing a resource for clinical translational research.

Journal of clinical and translational science·2017
Same author

Decision Support, Knowledge Representation and Management.

Yearbook of medical informatics·2016
Same journal

Digital divide in clinical and operational artificial intelligence adoption and implementation stages: US hospital diffusion patterns and AI deserts.

Journal of the American Medical Informatics Association : JAMIA·2026
Same journal

Extending the fundamental theorem of biomedical informatics: a proposal and illustrative examples.

Journal of the American Medical Informatics Association : JAMIA·2026
Same journal

Human factors methods for designing safe health information technology: what do the experts think?

Journal of the American Medical Informatics Association : JAMIA·2026
Same journal

Equity-by-design for socially assistive robots as digital health tools.

Journal of the American Medical Informatics Association : JAMIA·2026
Same journal

Orchestrator multi-agent clinical decision support system for secondary headache diagnosis in primary care.

Journal of the American Medical Informatics Association : JAMIA·2026
Same journal

CUI-Curate: a GraphRAG-based framework for automated clinical concept curation for NLP applications.

Journal of the American Medical Informatics Association : JAMIA·2026
See all related articles

Physician raters achieved high reliability when identifying clinical conditions from chest radiograph reports. A reference standard for information extraction can be reliably created with few raters for aggregate analysis, but more are needed for case-by-case assessment.

Area of Science:

  • Medical informatics
  • Natural language processing
  • Radiology

Background:

  • Establishing reliable reference standards is crucial for evaluating information extraction systems in healthcare.
  • Physician interpretation of clinical data from medical reports presents unique challenges for automated systems.

Purpose of the Study:

  • To assess the reliability of a reference standard for an information extraction task using physician raters.
  • To determine the number of raters required to achieve a reliable reference standard for different levels of analysis.

Main Methods:

  • Twenty-four physician raters from diverse sites and specialties evaluated clinical conditions based on chest radiograph reports.
  • Variance components and generalizability coefficients were calculated to estimate inter-rater reliability.

Related Experiment Videos

  • The number of raters needed for reliable reference standard generation was estimated.

    • Main Results:

    • Average per-rater reliability was 0.80, with individual condition reliability ranging from 0.67 to 0.97.
    • Central line presence and pneumothorax showed the highest reliability; pleural effusion and pneumonia showed the lowest.
    • One to two raters achieved 0.70 reliability, while six raters were needed for 0.95 reliability.

    Conclusions:

    • Physician raters demonstrate high reliability in identifying clinical conditions from text-based reports.
    • A single rater can create a reliable reference standard for aggregate system assessment, while six are needed for case-by-case evaluation.
    • These findings inform the design of future information extraction studies for natural language processing and knowledge-based systems.