Question 4.3 b) of the HW asked us to derive LR's for the BNP values at each given level in the provided table. The "hint" states that we should use "the general formula for LR's, not Sensitivity/(1-specificity)."
Does anyone know to what this "general formula" refers? I thought perhaps using a 2x2 table set-up, using a,b,c, and d, but that simplifies to Sensitivity/(1-specificity) - I thought that that's the whole point!
Anyone have some relief?!
In a test with more than two possible results, you should not calculate the likelihood ratio of a positive result, LR (+), or the likelihood ratio of a negative result, LR (-); you should calculate likelihood ratios LR(r) for each possible result r of the test. in the context of the problem on BNP, "r" refers to a BNP result that falls into a particular interval, like 80 - 150.
The general formula is:
LR (r) = P(r|D+)/P(r|D-)
This is the probability of result "r" in a patient With the disease Over the probability of result "r" in a patient WithOut the disease. (WOWO)
Instructions on how to calculate an interval likelihood ratio using data like those in Problem 4.3 were given in the lecture on 10/2 and are also in the answer to 4.3 B) at the back of the textbook.
