The Area Under the Curve (AUC) is a metric used to evaluate the performance of binary classification models, measuring the discrimination power between true positive and false positive rates [2] [3] [5] [8]. It can be calculated manually for a logistic regression model using a validation dataset with true dependent variable values and predicted retention status [7].

In the context of carboplatin dosing, AUC represents the area under the free carboplatin plasma concentration versus time curve. A carboplatin dosing calculator may use the Calvert method to calculate the total carboplatin dose needed to achieve a given AUC while taking into account renal function [1] [4].

To construct the ROC curve, calculate the true positive rate (TPR) and false positive rate (FPR) at every possible threshold, then graph TPR over FPR [5]. The AUC is calculated as the area below the ROC curve. A perfect model would have an AUC of 1.0, while a completely random model would have an AUC of 0.5 [3] [5].