changing null hypothesis value for sample size calculation of a single AUC

Question

I am trying to get the sample size for a single AUC using power.roc.test in the pROC package. For example, the expected AUC is 0.97 (alternative hypothesis) and the value I am comparing to is 0.95 (null hypothesis). At the significance level of 0.05 and power of 0.80, I get 433 positives and 433 negatives using MedCalc, a statistical software. However, I want to carry this out in R. I cannot find any package that allows me to set the null hypothesis value.

Does anyone know how to do this in R?

In pROC, I can use power.roc.test but there is no argument to set the null hypothesis value; it defaults to 0.50.

Answer 1

I have the exact same question. It is possible to compute a sample size using with two roc objects with power.roc.test(), e.g.

power.roc.test(roc1, roc2, power = 0.8)

although I want to compute a sample size just for two different AUC values (one representing the null hypothesis, like MedCalc).

Answer 2

I could also not find a solution in R. But if this helps, I found out that in MedCalc software you can define the null hypothesis value, e.g., as = 0.95. This is displayed e.g., here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6683590/