Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 63 views

    Conditional Probability and Medical Tests

    Uploaded by

    Ruslan Malovanyy
    Screening tests are used to find disease early in asymptomatic patients, while diagnostic tests confirm disease presence in symptomatic patients or screening positives. Medical tests are evaluated by comparing results to a gold standard, with statistics like sensitivity, specificity, and predictive values measuring accuracy. An example shows mammography is a good breast cancer screening test with high sensitivity and specificity, though its positive predictive value is low due to breast cancer's rare prevalence.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd

    Conditional Probability and Medical Tests

    Uploaded by

    Ruslan Malovanyy
    63 views20 pages
    Screening tests are used to find disease early in asymptomatic patients, while diagnostic tests confirm disease presence in symptomatic patients or screening positives. Medical tests are evaluated by comparing results to a gold standard, with statistics like sensitivity, specificity, and predictive values measuring accuracy. An example shows mammography is a good breast cancer screening test with high sensitivity and specificity, though its positive predictive value is low due to breast cancer's rare prevalence.

    Original Title

    Conditional+Probability+and+Medical+Tests (3)

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Screening tests are used to find disease early in asymptomatic patients, while diagnostic tests confirm disease presence in symptomatic patients or screening positives. Medical tests are evaluated by comparing results to a gold standard, with statistics like sensitivity, specificity, and predictive values measuring accuracy. An example shows mammography is a good breast cancer screening test with high sensitivity and specificity, though its positive predictive value is low due to breast cancer's rare prevalence.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    63 views20 pages

    Conditional Probability and Medical Tests

    Uploaded by

    Ruslan Malovanyy
    Screening tests are used to find disease early in asymptomatic patients, while diagnostic tests confirm disease presence in symptomatic patients or screening positives. Medical tests are evaluated by comparing results to a gold standard, with statistics like sensitivity, specificity, and predictive values measuring accuracy. An example shows mammography is a good breast cancer screening test with high sensitivity and specificity, though its positive predictive value is low due to breast cancer's rare prevalence.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    You are on page 1of 20

    Conditional Probability:

    Screening and Diagnostic


    Testing
    Types of Medical Tests
    • Screening tests are used to find people early in their disease
    progression
    • Used on asymptomatic patients
    • Diagnostic tests are used to confirm the presence or absence of
    disease
    • Used on symptomatic patients and/or patients who tested positive on
    screening tests
    Evaluating Medical Tests
    • No medical test is perfect
    • We evaluate the accuracy of medical tests by comparing it to a gold
    standard
    Evaluating Medical Tests

    Disease Present Disease Absent


    Test + True Positive False Positive
    Test - False Negative True Negative

    • True Positive: The patient has the disease and the test is positive
    • False Positive: The patient does not have the disease but the test is positive.
    • True Negative: The patient does not have the disease and and the test is negative
    • False Negative: the patient has the disease but the test is negative
    Example

    Disease Present Disease Absent


    Test + 312 6,976
    Test - 39 92,673

    • This table shows the results of a study evaluating a new diagnostic


    test for some disease.
    • How should we measure the accuracy of this test?
    • In other words, how can we tell if this is a good test to use in clinical
    practice?
    Evaluating Medical Tests
    • Sensitivity: the probability an individual tests positive given they have
    the disease.
    • Sensitivity = P(Test + | Disease +)
    • Refers to the ability of a test to correctly identify those patients with the
    disease
    • Specificity: the probability an individual tests negative given they do
    not have the disease.
    • Specificity = P(Test - | Disease -)
    • Refers to the ability of a test to correctly identify those patients without the
    disease
    Evaluating Medical Tests

    • Accuracy: The proportion of all tests that are correct.


    • Positive Predictive Value (PPV): the probability a patient has the
    disease given they test positive
    • PPV = P(Disease + | Test +)
    • How likely is it the patient has the disease given they test positive?
    • Negative Predictive Value (NPV): the probability a patient does not
    have the disease given they test negative
    • NPV = P(Disease - | Test -)
    • How likely is it the patient does not have the disease given they test negative?
    Evaluating Medical Tests
    • False Negative Fraction (FNF) = P(Test - | Disease +)
    • FNF = 1 – P(Test + | Disease +) = 1 - Sensitivity
    • False Positive Fraction (FPF) = P(Test + | Disease -)
    • FPF = 1 – P(Test - | Disease -) = 1 – Specificity
    • Likelihood Ratio+ (LR+) is defined as how much more likely it is that a
    patient who tests positive has the disease compared to one who
    doesn’t have the disease
    Summary of statistics that can be used to
    evaluate medical tests
    • Sensitivity = P(Test + | Disease +)
    • Specificity = P(Test - | Disease -)
    • Positive Predictive Value (PPV) = P(Disease + | Test +)
    • Negative Predictive Value (NPV) = P(Disease - | Test -)
    • Accuracy=
    • Likelihood Ratio+
    Example
    Disease Present Disease Absent
    Test + 312 6,976
    Test - 39 92,673

    This example considers mammography screening for breast cancer. In Canada around 2020, the prevalence of
    breast cancer for women over 40 was 0.35%

    How sensitive is the test?

    Introductory Statistics for the Life and Biomedical Sciences by Julie Vu & David Harrington
    Example
    Disease Present Disease Absent
    Test + 312 6,976
    Test - 39 92,673

    How specific is the test?


    Example
    Disease Present Disease Absent
    Test + 312 6,976
    Test - 39 92,673

    What is the chance the screening test fails to detect a case of cancer? That is what is the probability of a false
    negative (FNF)?

    Alternatively,
    Example
    Disease Present Disease Absent
    Test + 312 6,976
    Test - 39 92,673

    What is the chance the screening detecting a case of cancer when it is not present? That is what is the probability
    of a false positive (FPF)?

    Alternatively,
    Example
    Disease Present Disease Absent
    Test + 312 6,976
    Test - 39 92,673

    Is this a good screening test?


    • Sensitivity=88.9%
    • Specificity=93%
    • Chance of false negative=11%
    • Chance of false positive=7%
    Example
    Disease Present Disease Absent
    Test + 312 6,976
    Test - 39 92,673

    If a patient tests positive, what is the chance they have cancer (PPV)?
    PPV and NPV depend on disease prevalence

    Banoo, S., Bell, D., Bossuyt, P., Herring, A., Mabey, D., Poole, F., ... & Ramsay, A. (2007). Evaluation of diagnostic tests for infectious
    diseases: general principles. Nature Reviews Microbiology, 5(11), S21-S31.
    PPV and NPV depend on disease prevalence
    Remember:

    Using Bayes’ Theorem, we can rewrite PPV as:


    Overall accuracy also depends on prevalence
    Disease Present Disease Absent
    Test + 312 6,976
    Test - 39 92,673

    What is the overall accuracy of the test?


    Example
    Disease Present Disease Absent
    Test + 312 6,976
    Test - 39 92,673

    Is this a good screening test?


    • Sensitivity=88.9%
    • Specificity=93%
    • Chance of false negative=11%
    • Chance of false positive=7%
    • PPV=4%
    • Prevalence = 351/99649=0.35%

    This is a good screening test. Sensitivity and specificity are reasonably high. PPV is low because breast cancer is
    very rare.
    Key Points
    • Ideally, we want specificity and sensitivity to be as close to 100% as
    possible.
    • In practice though, there is a trade-off. Increasing the sensitivity of a
    test often decreases the specificity.
    • Sensitivity and specificity are intrinsic to the test. They do not depend
    on the population being tested.
    • Positive predictive value, negative predictive value, and overall
    accuracy are not intrinsic to the test. They depend on disease
    prevalence in the population being tested.

    You might also like