Jove
Visualize
Contact Us

Related Experiment Videos

Maximizing sensitivity in medical diagnosis using biased minimax probability machine.

Kaizhu Huang1, Haiqin Yang, Irwin King

  • 1Information Technology Laboratory, Fujitsu Research and Development Center Co., Ltd., Beijing 100016, China. kzhuang@frdc.fujitsu.com

IEEE Transactions on Bio-Medical Engineering
|May 12, 2006
PubMed
Summary

This study introduces the biased minimax probability machine (BMPM) for medical diagnosis. The BMPM improves sensitivity for detecting illness while maintaining specificity, outperforming traditional classifiers on real-world datasets.

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

SCMBench: benchmarking domain-specific and foundation models for single-cell multi-omics data integration.

Nature communications·2026
Same author

A Survey on Vision--Language--Action Models for Embodied AI.

IEEE transactions on neural networks and learning systems·2026
Same author

HC-GLAD: Dual hyperbolic contrastive learning for unsupervised graph-level anomaly detection.

Neural networks : the official journal of the International Neural Network Society·2026
Same author

Recent Advances of Multimodal Continual Learning: A Comprehensive Survey.

IEEE transactions on neural networks and learning systems·2026
Same author

Discrete Tokenization for Multimodal LLMs: A Comprehensive Survey.

IEEE transactions on pattern analysis and machine intelligence·2026
Same author

Learning Optimal Policies With Local Observations for Cooperative Multiagent Reinforcement Learning.

IEEE transactions on neural networks and learning systems·2026
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

Area of Science:

  • Machine Learning
  • Medical Diagnosis
  • Biomedical Informatics

Background:

  • Medical diagnosis using machine learning requires bias towards the 'ill' class to prevent delayed treatment.
  • Current methods for bias imposition are indirect and their systematic improvement is uncertain.
  • Improving sensitivity while maintaining specificity is crucial for effective medical diagnosis.

Purpose of the Study:

  • To introduce a novel learning tool, the biased minimax probability machine (BMPM), for direct and elegant bias imposition in medical diagnosis.
  • To achieve a bias favoring the 'ill' class by directly controlling worst-case accuracies.
  • To develop a distribution-free decision rule that maximizes worst-case sensitivity while ensuring acceptable worst-case specificity.

Main Methods:

Related Experiment Videos

  • The biased minimax probability machine (BMPM) was developed to directly control worst-case accuracies.
  • A distribution-free decision rule was derived to maximize worst-case sensitivity and maintain worst-case specificity.
  • The BMPM's performance was evaluated against k-nearest neighbor, naive Bayesian, and C4.5 classifiers.

    • Main Results:

    • The BMPM directly incorporates bias toward the 'ill' class by controlling worst-case accuracies.
    • The BMPM provides a distribution-free decision rule, distinguishing it from generative classifiers.
    • On breast-cancer and heart disease datasets, the BMPM demonstrated superior performance compared to traditional classifiers.

    Conclusions:

    • The BMPM offers a more rigorous and direct approach to handling medical diagnosis bias.
    • The distribution-free nature of the BMPM enhances its applicability across various datasets.
    • The BMPM represents a significant advancement in machine learning for medical diagnosis, outperforming existing methods.