| Measure | This run | Expected |
|---|
Simplifications: each person's true status is drawn by chance from the prevalence, and each test result by chance from the sensitivity and specificity, independently. Real screens report risk scores with cutoffs, can give no-call results, and their performance varies by group. The diagnostic follow-up test is set at 99.5% sensitivity and 99.95% specificity (teaching values). Preset numbers are rounded teaching values, not figures for any specific product.
Problems are randomly generated with fictional people and teaching numbers. Percentages can be typed as 33, 33% or 0.33. Predictive values are accepted within about half a percentage point (NPV within 0.15); counts within 1 or 1 percent. Gene names are real, but variants and reports are invented for practice.
Your job
Drag a card onto a box, or tap a card and then tap the box's name. Cards are buttons, so Tab and Enter work too.
Fictional people and teaching data. For learning only, not medical advice or genetic counseling. Real decisions involve much more information and a qualified clinician or genetic counselor.
Wrong picks are fine here. Each one tells you why.