How would you select the right metric, such as Precision, Recall, F1, or ROC-AUC, for an imbalanced healthcare classification problem?
