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

Properties of diagnostic data distributions

A P Dawid

    Biometrics
    |September 1, 1976
    PubMed
    Summary

    This study proposes focusing on "diagnostic distributions" for medical diagnosis, simplifying statistical analysis by adjusting for selection bias and population changes. This approach enhances the accuracy of diagnosing new patients using symptom data.

    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

    Still further remarks on: "Paternity analysis in special fatherless case without direct testing of alleged father" [Forensic Science International 146S (2004) S159-S161] and remarks on it [FSI 163 (2006) 158-160, FSI 172 (2007) e6-e8].

    Forensic science international·2011
    Same author

    Estimating mutation rates from paternity casework.

    Forensic science international. Genetics·2008
    Same author

    Object-oriented Bayesian networks for complex forensic DNA profiling problems.

    Forensic science international·2006
    Same author

    Remarks on: "Paternity analysis in special fatherless cases without direct testing of alleged father" [Forensic Science International 146S (2004) S159-S161].

    Forensic science international·2005
    Same author

    A statistical treatment of biases affecting the estimation of mutation rates.

    Mutation research·2004
    Same author

    Probabilistic expert systems for DNA mixture profiling.

    Theoretical population biology·2003

    Area of Science:

    • Statistics
    • Medical Diagnosis
    • Biostatistics

    Background:

    • Statistical methods are crucial for medical diagnosis, utilizing patient disease and symptom databases.
    • Data complexity arises from factors like selection bias and unstable populations.
    • Current diagnostic statistics often focus on symptom incidence for a given disease, which may be suboptimal.

    Purpose of the Study:

    • To re-evaluate statistical approaches in medical diagnosis.
    • To address challenges posed by selection bias and unstable populations in diagnostic databases.
    • To advocate for a shift towards studying "diagnostic distributions" as the primary focus.

    Main Methods:

    • Analysis of statistical methods for medical diagnosis considering selection bias and population dynamics.
    • Mathematical modeling to understand the transformation of diagnostic distributions under changing population structures.
    • Evaluation of the generalized logistic model within the proposed diagnostic distribution framework.

    Main Results:

    • Correction for selection bias is often unnecessary when assessing disease probabilities from symptom information.
    • Diagnostic distributions exhibit a simple transformation form under many changing population structures.
    • The proposed focus on diagnostic distributions offers a more appropriate framework than the traditional incidence-based approach.

    Conclusions:

    • The study recommends prioritizing "diagnostic distributions" over traditional symptom incidence models in medical statistics.
    • This shift can simplify analysis and improve diagnostic accuracy, especially with evolving population demographics.
    • The generalized logistic model is well-suited for this enhanced diagnostic statistics framework.

    Related Experiment Videos