Left: test results and the 2×2 table in a population of 400, pre- and post-test diseased counts in a 100-person grid, and a threshold moving over the healthy and diseased distributions. Right: PPV and NPV against prevalence, pre- versus post-test probability for several likelihood ratios, and the ROC curve.

Continuous stream · no seeking

Sensitivity, specificity and likelihood ratios belong to the test; what a result means for a patient depends on the pre-test probability (prevalence) and the chosen threshold.

Source: formulas and interpretation thresholds from the book (29.1-29.2); Sn/Sp examples from the book (Figure 29.1, question b29:1). ROC calculated with an equal-variance binormal model (AUC = Φ(d′/√2)).

How to use: a single click on the drawing pauses and resumes; ↑ ↓ or the buttons below change the test, the likelihood ratio or the AUC; use the time bar below to set the prevalence or the threshold where you want it.

Did this animation help you understand the topic?

Diagnostic test performance · Open in Nuclear Medicine Explained ↗