MSBI: how to create a measure with DAX that implements statistical de-identification?

I need to apply statistical de-identification rules in an MSBI Tabular model. An example of the kinds of rules I'm trying to implement is: if N < 20,000, then any row less than 11 must be combined with other rows. Example: I report counts of Foo by Dimension Bar. All these issues are discussed in Msbi Certification. Since the total (in this ALLSELECTED context) is less than 20K Foo, this output is not valid:

Bar 1: 500

Bar 2: 400

Bar 3: 100

Bar 4: 11

Total: 1011

The valid output would be valid:

Bar 1: 500

Bar 2: 400

Bar 3 & 4: 111

Total: 1011

I want to use a measure so that when users change from Dimension Bar to Dimension Baz, or when I change the slicers being used, I want these rules to recalculate on the fly - so just combining the rows for that one dimension in the model is not the solution I'm looking for. Thanks!