Can sensitivity be calculated in non-binary variables?

Hi everyone,

Is there any analogous measure of sensitivity and specificity for non-binary variables?

For example, if I want to study the sensitivity of a test that analyzes blood types (i.e. A, B, AB or 0) in comparison with a gold standard tet. What's the best way to do it?

I have thought about doing 4 contingency tables, and in each one of them I can get a value of the test for that blood type. However, how can I translate the findings of each category to the test as a whole? I guess I can simply calculate the average sensitivity value, but is there any other possibility?

Thanks in advance.