Positive and negative predictive values


Positive predictive valuePositive predictive value (PPV) is the probability that subjects with a positive screening test truly have the disease.
and
negative predictive valueNegative predictive value (NPV) is the probability that subjects with a negative screening test truly don't have the disease.

are statistics that quantify the clinical usefulness of
prognostic and diagnostic medical testsA medical test is a medical procedure performed to predict, diagnose, or monitor a condition.
in
low prevalent diseasesPrevalence is the proportion of a population found to have a condition.
.

This educational software tool allows to explore and visualise the predictive performance of
medical testsA medical test is a medical procedure performed to predict, diagnose, or monitor a condition.
by visualising the following parameters:
-
area under the ROC curveThe area under a ROC curve quantifies the overall ability of a test to discriminate between those individuals with the disease and those without the disease.
(
AUROCThe area under a ROC curve quantifies the overall ability of a test to discriminate between those individuals with the disease and those without the disease.
),
-
sensitivityTest sensitivity is the ability of a test to correctly identify those with the disease (true positive rate).
and
specificityTest specificity is the ability of the test to correctly identify those without the disease (true negative rate).
,
-
prevalence of the diseasePrevalence is the proportion of a population found to have a condition.
,
-
PPVPositive predictive value (PPV) is the probability that subjects with a positive screening test truly have the disease.
and
NPVNegative predictive value (NPV) is the probability that subjects with a negative screening test truly don't have the disease.
,
- number of samples assayed,
- distribution of the test score (randomly generated using user-defined
AUROCThe area under a ROC curve quantifies the overall ability of a test to discriminate between those individuals with the disease and those without the disease.
,
skewnessSkewness is a measure of the asymmetry of the probability distribution of a random variable.
and
random seedA random seed is a number used to initialize a random number generator.
).

Note that the number of samples assayed to estimate the predictive performance of a
clinical testA medical test is a medical procedure performed to predict, diagnose, or monitor a condition.
does not always reflect the
prevalence of the diseasePrevalence is the proportion of a population found to have a condition.
. Such experimental designs allow to decrease the cost of experiments in the case of
low prevalent diseasesPrevalence is the proportion of a population found to have a condition.
. In such cases, the computation of statistics such as
PPVPositive predictive value (PPV) is the probability that subjects with a positive screening test truly have the disease.
and
NPVNegative predictive value (NPV) is the probability that subjects with a negative screening test truly don't have the disease.
must take into account
disease prevalencePrevalence is the proportion of a population found to have a condition.
. The present tool allows to explore sampling size independently from
disease prevalencePrevalence is the proportion of a population found to have a condition.
.

The R package alternativeROC provides functions for computation of ROC curves, PPV, NPV and equi-predictive lines.
 

Copyright © 2015-2017 SQU4RE bvba