Q: "Is it possible for Alice to know which particle has been measured on the other side, and which didn't?"
A: No, assuming you want to do that without using a classical signal - i.e. you wish Alice to examine just one of the two entangled particles. The reason is exactly as you say: outcomes are random, and there isn't much useful information to be found in a random value.
You also said: "I'm assuming unmeasured particles with superposition can show an interference pattern and measured ones don't." And as you might be thinking about now, that assumption is not valid. The reason: entangled particles do not produce interference patterns in the manner that unentangled particles might. Please see this reference by 2022 Nobel laureate Anton Zeilinger, figure 2 and also formula (4). Because they do not produce an interference pattern, collapsing one will not change anything on the other side - even if you look at a large number of such pairs.
https://courses.washington.edu/ega/more_papers/zeilinger.pdf
Also, it is worth mentioning: The order of "collapse" of entangled particles never makes any difference to the observed statistical results. And in fact one should not think of cessation of entanglement as being "caused" by one or the other. It is not meaningful to assign a time to the collapse action (assuming it is a physical action at all, which is not generally accepted one way or the other).