I saw that you have calculated the absolute difference between suspects and bots... Here are the relative differences (if difference is in the +-5% intervall the cell is red)...
there are indeed some interesting similarities.
You should look more into it and also compare the stats by position and not only overall.
The problem is that only the room/network itself can really check and investigate what is going on.
But suspicous this is for sure. But maybe this is a studygroup and the players play all same way. Who knows?
What are the EV-winrates of those guys buy the way?
Don`t tell me that they are loosing...