My idea would be to use the Matcher as @takashi said, paired with a NeighborFinder. Set the Matcher to Check Attributes using your ID attribute. With the NeighborFinder set to Candidates Only for the input you can connect just the Matched output from Matcher. You can then use Group Processing, Group By the _match_id attribute created by the matcher, and make sure to set Treat Polygons As to Areas.
The Matcher will tell you which features have the same ID, and then sending those features through the NeighborFinder will tell you which ones are disjoint as they should have a _distance value greater than 0.