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

Combining predictors for classification using the area under the receiver operating characteristic curve.

Margaret Sullivan Pepe1, Tianxi Cai, Gary Longton

  • 1Program in Biostatistics and Biomathematics, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue N., M2-B500, Seattle, Washington 98109-1024, USA. mspepe@u.washington.edu

Biometrics
|March 18, 2006
PubMed
Summary

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

Cancer risk with methotrexate-TNF antagonist combination therapy compared to TNF-antagonist monotherapy in IBD.

Inflammatory bowel diseases·2026
Same author

Toward Artificial Intelligence-driven Clinical Decision Support Tools in Rheumatology.

Rheumatic diseases clinics of North America·2026
Same author

Statistics and AI - A Fireside Conversation.

Harvard data science review·2026
Same author

Predicting the timing of first sustained cognitive worsening in Alzheimer's disease using real-world clinical data and machine learning.

medRxiv : the preprint server for health sciences·2026
Same author

Nonparametric estimation of the total treatment effect with multiple outcomes in the presence of terminal events.

Biometrics·2026
Same author

Stratification of Alzheimer's disease patients using knowledge-guided unsupervised latent factor clustering with electronic health record data.

Communications medicine·2026

Combining cancer biomarkers using the area under the curve (AUC) offers a robust alternative to traditional likelihood methods. This approach provides accurate classification and prediction, especially when standard models do not fit the data.

Area of Science:

  • Biostatistics
  • Biomarker Discovery
  • Machine Learning for Healthcare

Background:

  • Single cancer biomarkers lack sufficient sensitivity and specificity for effective screening.
  • Combining multiple biomarkers is necessary for accurate cancer classification.
  • Traditional methods often optimize the likelihood function for marker combination.

Purpose of the Study:

  • To evaluate the area under the empirical receiver operating characteristic curve (AUC) as an alternative objective function for combining biomarkers.
  • To compare the performance of AUC-based marker combination scores with traditional likelihood-based methods.

Main Methods:

  • Utilized generalized linear models for risk score estimation.
  • Employed area under the empirical receiver operating characteristic curve (AUC) as the objective function.

Related Experiment Videos

  • Conducted simulation studies and analyzed proteomics biomarker data.

    • Main Results:

    • AUC-based methods provide consistent parameter estimates without requiring a specified link function.
    • AUC-based classification scores perform comparably to logistic likelihood-based scores when the logistic model is appropriate.
    • AUC-based scores significantly outperform logistic regression scores when the logistic model assumption is violated.

    Conclusions:

    • Maximizing AUC is a viable and potentially superior strategy for deriving marker combination scores for classification and prediction.
    • This method offers improved performance in biomarker studies, particularly when underlying data distributions deviate from standard logistic models.