It definitely seems like some simple heuristics could help a lot in giving police a sense of the probability that it's a false alarm. They will of course still need to assess the situation, but systemic police accountability issues aside, harming completely innocent people is still no picnic for the officers involved, so knowing when the fraud likelihood is particularly high should help them calm down and take a more measured approach.

Apart from call tracing, I'd expect simple things like whether the caller stays on the line and cooperates with the dispatcher (vs hanging up immediately) to be quite predictive.