Biomarker Tools
This toolset estimates risk stratification from early biomarker data and provides strategies to advance biomarkers or other risk
measures identified through casecontrol studies to clinical or public health applications. The toolset will show quantities
for which people's intuition is poor, such as need for a single marker of a rare disease to improve management by some serious
intervention. We hope these strategies will help researchers to identify and promote the most promising markers early on,
illuminate any necessary improvements, and eliminate markers that are most likely to fail.
From Difference in Means to Risk Stratification: A Web Tool
Calculating
Biomarker Comparison
This tool plots contours of likelihood ratio positive (LR+) and
likelihood ratio negative (LR) for a reference test with indicated
sensitivity and specificity pairs (see the example plot below). The
likelihood ratio contours define four areas:

Area A shows combinations of
sensitivity and specificity with higher LR+ and LR than the
reference test. This corresponds to a higher positive predictive
value (PPV) and lower complement of the negative predictive value
(cNPV)

Area B shows combinations of
sensitivity and specificity with higher LR+ and lower LR than
the reference test. This corresponds to a higher positive
predictive value (PPV) and higher complement of the negative
predictive value (cNPV).

Area C shows combinations of sensitivity and specificity
with lower LR+ and higher LR compared to the reference
test. This corresponds to a lower positive
predictive value (PPV) and lower complement of the
negative predictive value (cNPV).

Area D shows combinations of sensitivity and
specificity with lower LR+ and lower LR compared
to the reference test. This corresponds to a lower positive
predictive value (PPV) and higher complement of the negative
predictive value (cNPV)
LR+ and LR measures provide testspecific characteristics of risk
stratification that yield estimates of absolute risk (PPV and NPV)
when multiplied with the specific disease prevalence. LR+ and LR
estimated in one population will be the same in another
population, whenever sensitivities and specificities are the same
in the 2 populations, even when disease prevalences are much
different. To calculate PPV and cNPV, provide a prevalence value.
References:
Marina V. Kondratovich, (2007),
Comparing Two Medical Tests
When Results of Reference Standard Are Unavailable for Those
Negative via Both Tests, Journal of Biopharmaceutical Statistics, 18:1, 145166,DOI: 10.1080/10543400701668308
Risk Stratification Advanced Analysis
Help
Tools Help
Risk Stratification Advanced Analysis
Instructions for Input
This tool calculates values and creates graphs for valid combinations of given PPV, cNPV,
delta, specificity, sensitivity, and prevalence values. Click on the 'Example' link
for sample input combinations.

Choose values for the
independent variable, or for the xaxis of the output graph, from
the dropdown menu next to 'Independent Variable'. Enter your
values as decimals separated by commas.

Choose values for the contours
of the output graph from the dropdown menu next to 'Contour'.
Enter your values as decimals separated by commas.

Choose fixed values.
Then enter your values as decimals separated by commas.
 Click 'Calculate'.
Input Validation Rules
 Specificity, Sensitivity, PPV, cNPV, and Prevalence can only be 0 to 1
 Delta can be 0 to 5
 cNPV < Prevalence
 For arrays: max(cNPV) < min(Prevalence)
 Prevalence < PPV
 For arrays: max(prev) < min(PPV)
 Sensitivity+Specificity1 > 0

PPV and cNPV; Sensitivity, Specificity, and Delta; PPV,
Prevalence, and Delta; and cNPV, Prevalence, and Delta are invalid combinations of input
FAQ
 Where do I go for technical support?
Please send an
email to our technical support team.
 What browsers do the web tools support?
The web tools have been designed and tested with Internet Explorer 10, Firefox and Chrome.
They do not support Internet Explorer 9 and below.