Reporting only sensitivity/specificity/ROC metrics and not reporting precision/positive predictive value is a bad sign. Especially since the latter is what heal

Reporting only sensitivity/specificity/ROC metrics and not reporting precision/positive predictive value is a bad sign. Especially since the latter is what health systems will want to look at before deciding on implementation. The fact that they fiddled with the balance of classes in the test set makes the above even worse.