Area Under the Curve (AUC)

In pharmacokinetics, AUC is commonly estimated from measured concentration points using the trapezoidal rule, then extended beyond the last sample by estimating the terminal elimination slope for AUC0-inf. In ROC analysis, AUC can be interpreted probabilistically: the chance that a randomly chosen positive case receives a higher score than a randomly chosen negative case.

The same three letters can mean two very different things

AUC creates a very specific trap: in one paper it means how much of a substance the body saw over time, and in another it means how well a test separates true cases from false alarms. Same acronym, same phrase, completely different question. That is why people end up asking things like What does an AUC of 0.7 mean? without first asking, Which curve are we talking about?

The picture to keep in your head

Imagine a hill drawn on graph paper. Counting only the peak tells you how tall the hill got for one instant. Counting the land under the hill tells you how much landscape is really there.

That is the surprise of Area Under the Curve. AUC is not about the single highest point. It is about the total accumulation under a plotted line.

In pharmacokinetics—the study of what the body does to a drug or supplement—AUC usually means the area under the concentration-time curve. On the x-axis is time; on the y-axis is the amount in blood. A bigger AUC usually means greater overall exposure. Two products can hit different peaks yet still have similar AUCs if the total exposure across hours is similar. That is why regulators use AUC in bioavailability and bioequivalence work.

In ROC AUC, the curve is different. The x-axis is false positives; the y-axis is true positives. Here AUC tells you how well a model or test ranks real signal above noise across all possible cutoffs, not just one chosen threshold. In this setting, 0.5 suggests chance-level discrimination—basically coin-flip performance. 0.7 usually means fair discrimination: better than chance, useful in some contexts, but far from excellent. Very roughly, 0.8 is often called good and 0.9 excellent, though the labels depend on context and the cost of being wrong.

Why PR AUC exists

Now another twist: when positive cases are rare, ROC curve summaries can look more flattering than the real-world task feels. That is why researchers sometimes report PR AUC—area under the precision-recall curve—which pays much more attention to performance on the rare positive class. So a “good AUC score” is not a universal badge. It depends on which curve, which baseline, and what kind of mistake hurts more.

One decision that helps immediately

When you see AUC in a supplement or drug paper, make one decision first: identify the axes before interpreting the number. If it is a concentration-time graph, think total exposure. If it is ROC AUC or PR AUC, think sorting ability across thresholds. That single move prevents the most common category error and makes the rest of the paper much easier to read.

A few label and paper conventions help: AUC0-t means area from time zero to the last measured point; AUC0-inf means the measured area plus the estimated tail out to infinity; AUROC is another name for ROC AUC.