Fig. 2

ROC curve comparing NAAT CT value vs toxin EIA positivity. Note: The area under the ROC curve (AUC) was equal to AUC = 0.819. The y-axis is the True Positive Fraction, or sensitivity, and the x-axis is the False Positive Fraction, which is equal to 1 − specificity. The True Positive Fraction is the probability that for a fixed cut off, c, the classifier (M) gives a positive result when the true result (D) is positive, i.e. TPF(c) = P(M ≥ c| D = Positive). In a similar fashion, the False Positive Fraction is the probability that for a fixed cut off the classifier gives a positive result when the true result is negative, i.e. TPF(c) = P(M ≥ c| D = Negative)