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

An improved measure for comparing diagnostic tests.

N M Adams1, D J Hand

  • 1Department of Mathematics, Imperial College, London, UK. n.adams@ic.ac.uk

Computers in Biology and Medicine
|March 14, 2000
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

Quadruplex priming amplification for the detection of mRNA from surrogate patient samples.

The Analyst·2014
Same author

Interaction of light and temperature on seed germination of Rumex obtusifolius L.

Planta·2013
Same author

Mining medical data.

Statistical methods in medical research·2000
Same author

Improving the practice of classifier performance assessment.

Neural computation·2000
Same author

Clinical impact versus factor analysis for quality of life questionnaire construction.

Journal of clinical epidemiology·1998
Same author

Generalisation from phase III clinical trials: survival, quality of life, and health economics.

Lancet (London, England)·1997
Same journal

A computational model of chemically- and mechanically-induced thrombus formation in cerebral aneurysms.

Computers in biology and medicine·2026
Same journal

An improved catch fish optimization based deep learning model for Parkinson disease classification using EEG signal.

Computers in biology and medicine·2026
Same journal

Assessing the robustness of evaluation metrics for synthetic ECG signal quality.

Computers in biology and medicine·2026
Same journal

Integrating stemness and epithelial-mesenchymal transition signatures with machine learning identifies RUNX1 as a therapeutic vulnerability in colorectal cancer.

Computers in biology and medicine·2026
Same journal

Differential regional textural attributes of tongue in normal and acidity patients in the light of traditional Chinese medicine.

Computers in biology and medicine·2026
Same journal

SC-MSDNet: Spatial-consistent multi-view self-distillation for retinal OCT classification.

Computers in biology and medicine·2026
See all related articles

This study introduces a novel loss-based method for evaluating diagnostic tests, incorporating misclassification costs. This approach offers a more accurate comparison of predictive performance than traditional metrics, preventing potentially flawed conclusions.

Area of Science:

  • Medical diagnostics
  • Decision analysis
  • Statistical modeling

Background:

  • Standard methods for assessing diagnostic test performance, such as the area under the receiver-operating characteristic curve (AUC) and misclassification rate, do not inherently account for the costs associated with incorrect diagnoses.
  • Ignoring misclassification costs can lead to suboptimal decision-making in clinical practice, potentially resulting in patient harm or inefficient resource allocation.

Purpose of the Study:

  • To introduce and validate a novel loss-based framework for comparing the predictive performance of diagnostic tests.
  • To demonstrate the limitations of traditional performance metrics when misclassification costs are relevant.
  • To provide a more accurate and cost-sensitive method for diagnostic test evaluation.

Main Methods:

Related Experiment Videos

  • Development of a loss function that integrates diagnostic test predictions with associated misclassification costs.
  • Comparison of the proposed loss-based method against standard metrics (e.g., AUC, misclassification rate) using simulated and real-world data.
  • Illustrative examples to highlight the impact of incorporating cost information.

    • Main Results:

    • The proposed loss-based method provides a more nuanced evaluation of diagnostic test performance by explicitly considering misclassification costs.
    • In scenarios with significant cost differentials, standard metrics can yield misleading conclusions regarding the optimal test or decision threshold.
    • The method demonstrates its utility in identifying the most cost-effective diagnostic strategies.

    Conclusions:

    • A loss-based approach is superior to standard metrics for evaluating diagnostic tests when misclassification costs are known or can be estimated.
    • Incorporating economic factors into performance assessment is crucial for informed clinical decision-making and resource optimization.
    • This framework offers a valuable tool for researchers and clinicians to select and interpret diagnostic tests more effectively.