Chemical Pathology Math Flashcards
Suggest the formula for Positive Predictive Value
Suggest the formula for Negative Predictive Value
Suggest the formula for Positive Predictive Value
TP/(TP+FP)
Denominator: All ppl tested positive
Suggest the formula for Negative Predictive Value
TN/(TN+FN)
Denominator: All ppl tested negative
Suggest the formula for Sensitivity
Suggest the formula for Specificity
Suggest the formula for Sensitivity
TP/(TP+FN)
Denominator: All ppl actually have the disease
Suggest the formula for Specificity
TN/(TN+FP)
Denominator: All ppl NOT have the disease
Relationship between Threshold and Sensitivity and Specificity (REFER TO DIAGRAM on notes)
Fill in the blanks with TP, FP, TN, FN
The left curve is normal; the right curve is diseased.
At point P or rightwards, there is only __________, no more _______________.
At point K or leftwards, there is only ____________, no more _______________.
So when the threshold increases (moves from left to right), the ____________ increases.
So when the threshold decreases (moves from right to left), the _____________ increases.
Relationship between Threshold and Sensitivity and Specificity (REFER TO DIAGRAM on notes)
Fill in the blanks with TP, FP, TN, FN
The left curve is normal; the right curve is diseased.
At point P or rightwards, there is only TRUE POSITIVE, no more FALSE POSITIVE.
At point K or leftwards, there is only TRUE NEGATIVE, no more FALSE NEGATIVE.
So when the threshold increases (moves from left to right), the SPECIFICITY increases.
So when the threshold decreases (moves from right to left), the SENSITIVITY increases.
Linking Sensitivity with NPV
When Sensitivity increases, how would the NPV change?
Sensitivity = TP/(TP+FN) NPV = TN/(TN+FN)
Sensitivity increases
FN decreases.
NPV INCREASES
Linking Specificity with PPV
When Specificity increases, how would the PPV change?
Specificity = TN/(TN+FP) PPV = TP/(TP+FP)
Specificity increases
FP decreases
PPV INCREASES
How to calculate the likelihood ratio of LR+
How to calculate the likelihood ratio of LR-
How to calculate the likelihood ratio of LR+
(Prob result is +ve in diseased ppl) /
(Prob result is +ve in non-diseased ppl)
P (T+ | D+) /
P (T+ | D-)
(Sensitivity) /
(1 - Specificity)
How to calculate the likelihood ratio of LR-
(Prob result is -ve in diseased ppl) /
(Prob result is -ve in non-diseased ppl)
P (T- | D+) /
P (T- | D-)
(1 - Sensitivity) /
(Specificity)
If the likelihood ratio for tests = 1
What does it mean
The test is useless
Cuz Sensitivity + Specificity = 1
Which is equal to flipping a coin
Consider a R-C curve (operating characteristic curve)
The Y-axis is ____________
The X-axis is ____________
When the area under the curve increases, what does it mean?
Consider a R-C curve (operating characteristic curve)
The Y-axis is (SENSITIVITY)
The X-axis is (1 - SPECIFICITY)
When the area under the curve increases, what does it mean?