Analytics have not come up very much. There was one example from a session this morning talking about the bomber incident at the Ariana Grande concert, where apparently the bomber was loitering in the area for a bit, and had they known about that, they might have been able to confront him. Of course, you don't always know where a person would be loitering, so you'd need lots of cameras, lots of analytics (and then by extension, you would get lots of alarms), so it is not really clear that video analytics would have done much in that case.
When the threats are often suicide bombers, people driving vehicles into crowds, or terrorist drones (so far more of a theoretical threat than an actual one), early detection and warning is more about proper screening, limiting access, keeping areas where crowds will be unaccessible by vehicles (bollards, etc.).
Another threat is rowdy crowds/fans reacting to situations, which analytics will also not help with.