Mean prevalence-related metrics for our 8 models across the 10 train-test splits.

Metrics included are false positives (FP); false negatives (FN); number of positive calls (Pos calls); number of true positives in the test set; discordance; and the difference in discordance from the least discordant model, the SVM. Here, discordance is the difference between the predicted percentage positive and the true percentage positive. Simultaneous confidence intervals are multiplicity-adjusted to control FWER.