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

High performance for expert systems: I. Escaping from the demonstrator class.

P L Alvey, C D Myers, M F Greaves

    Medical Informatics = Medecine Et Informatique
    |April 1, 1987
    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

    Inappropriate rearrangement of immunoglobulin and T-cell receptor genes.

    Immunology today·2014
    Same author

    Specificity of anti-theta sera in cytotoxicity and functional tests on T lymphocytes.

    Nature: New biology·2010
    Same author

    HLA-associated susceptibility to childhood B-cell precursor ALL: definition and role of HLA-DPB1 supertypes.

    British journal of cancer·2008
    Same author

    Massage for mechanical neck disorders.

    The Cochrane database of systematic reviews·2006
    Same author

    Cord blood donor cell leukemia in recipients.

    Leukemia·2006
    Same author

    Day care in infancy and risk of childhood acute lymphoblastic leukaemia: findings from UK case-control study.

    BMJ (Clinical research ed.)·2005

    Converting a demonstrator expert system for leukaemia diagnosis to a high-performance system requires addressing domain representation flaws. Correcting minor errors improved performance to 94%, but complex cases necessitate fundamental changes for accuracy.

    Area of Science:

    • Medical Informatics
    • Artificial Intelligence in Medicine

    Background:

    • Expert systems for leukaemia diagnosis face challenges in scaling from demonstrator to high-performance systems.
    • Limited domain representation and minor logical errors were identified in the demonstrator system.

    Purpose of the Study:

    • To investigate the problems encountered when converting a demonstrator expert system for leukaemia diagnosis into a large, high-performance system.
    • To assess the impact of system imperfections on diagnostic accuracy and performance.

    Main Methods:

    • Logical completeness analysis of the demonstrator expert system.
    • Evaluation of system performance on simple and complex diagnostic test cases.
    • Assessment of the role of default mechanisms and numerical uncertainty handling.

    Related Experiment Videos

    Main Results:

    • Correction of minor errors and limited domain representation issues improved system performance to 94% on simpler cases.
    • Complex cases revealed significant limitations, suggesting fundamental changes to domain representation are needed.
    • Default mechanisms concealed errors, and numerical uncertainty methods offered no added value.

    Conclusions:

    • High-performance systems require accurate domain representation; imperfections must be corrected before adding complexity.
    • Demonstrator systems can tolerate minor inaccuracies, but these become critical in high-performance applications.
    • Default mechanisms and specific uncertainty handling methods were found to be detrimental.