Sensitivity and specificity

I can never remember the definitions, so note to self:

Specificity: proportion of true negative result out of all negatives. Can it correctly detect someone without the disease? True negative rate.

Sensitivity: proportion of positive result given the person really is positive. Can it detect someone with a disease? True positive rate.

A very sensitive test won’t have many false negatives. A very specific test won’t have many false positives.

The nhsx AI Dictionary is good for definitions.