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

Improving the classification of multiple disorders with problem decomposition.

Radwan E Abdel-Aal1, Mona R E Abdel-Halim, Safa Abdel-Aal

  • 1Computer Engineering Department, King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia. radwan@kfupm.edu.sa <radwan@kfupm.edu.sa>

Journal of Biomedical Informatics
|January 31, 2006
PubMed
Summary

Differential diagnosis in clinical medicine is improved by decomposing complex problems into simpler ones. Hierarchical decomposition using abductive networks boosts diagnostic accuracy to 99% and reduces costs.

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

Peripilar Sign in Androgenetic Alopecia: Does It Really Indicate Peripilar Infiltrate?

Dermatology practical & conceptual·2024
Same author

Pigmented colloid milium in a beta-thalassemia major patient: a case report.

International journal of dermatology·2023
Same author

Postpartum Castleman disease presenting as paraneoplastic pemphigus: a case report.

International journal of dermatology·2022
Same author

Blinding Xanthoma Disseminatum.

American journal of ophthalmology·2022
Same author

Role of streptococcal infection in the etiopathogenesis of pityriasis lichenoides chronica and the therapeutic efficacy of azithromycin: a randomized controlled trial.

Archives of dermatological research·2022
Same author

Eczematous mucinous eccrine nevus: a novel presentation with Meyerson phenomenon.

International journal of dermatology·2022

Area of Science:

  • Artificial Intelligence in Medicine
  • Machine Learning for Diagnostics
  • Clinical Decision Support Systems

Background:

  • Differential diagnosis of multiple disorders presents a significant challenge in clinical medicine.
  • The divide-and-conquer principle offers a strategy to simplify complex diagnostic problems by breaking them into manageable sub-problems.
  • Abductive networks are a type of machine learning model suitable for classification tasks.

Purpose of the Study:

  • To investigate the effectiveness of problem decomposition strategies for differential diagnosis.
  • To evaluate the performance of abductive network classifiers using different decomposition approaches on a dermatology dataset.
  • To assess the impact of decomposition on classification accuracy, model complexity, and diagnostic insight.

Main Methods:

Related Experiment Videos

  • Applied abductive network classifiers to a 6-class standard dermatology dataset.
  • Investigated three problem decomposition scenarios: class decomposition and two hierarchical approaches (based on clinical practice and class separability).
  • Compared a two-stage classification scheme based on hierarchical decomposition with a single-classifier monolithic approach.

Main Results:

  • Hierarchical decomposition significantly boosted classification accuracy from 91% (monolithic approach) to 99%, reaching the theoretical upper limit.
  • Decomposed models reduced the number of required input variables by up to 47%, enhancing cost-effectiveness and convenience.
  • Automatic input selection by abductive networks provided insights into diagnostic problem structure and marker value.

Conclusions:

  • Problem decomposition, particularly hierarchical approaches, is a highly effective strategy for improving differential diagnosis accuracy.
  • This approach simplifies diagnostic models, reduces costs, and enhances clinical decision-making support.
  • The findings support the use of machine learning and problem decomposition for more efficient and insightful medical diagnostics.