Hi,
I am performing a biomarker study and I have a certain AUC value after performing machine learning algorithm (radom forest).
Is there a way to tell how many additional samples (from control and disease) I will need if I want to, say, increase AUC from 0.77 to 0.85? (regardless of the desired power and p-value)
A realted question is whether I can know how the AUC will increase with increasing sample size?
Thanks a lot
Iddo
The AUC value is a summary measure of the accuracy of a quantitative diagnostic test. If you obtained 0.77 using optimal sample size (obtained from a power test) then the AUC will not change with increased sample size. The only way then to change AUC is to use a different diagnostic test or algorithm :)
In geographic and environmental sciences the optimal sample size is reported as between 10 and 100; more is not better as Andrew also commented.
Maybe you want to have a look at the power.roc.test function of R's pROC package:
https://cran.r-project.org/web/packages/pROC/pROC.pdf
- Badges
- Science topic
- Similar topics
- Biological Science
- Microbiology
- Virology
- Virus