In all of these cases the "AI" is assessing some features of the patient scans and history to reach it's conclusions. Why isn't the method ("algorithm") made c

In all of these cases the "AI" is assessing some features of the patient scans and history to reach it's conclusions. Why isn't the method ("algorithm") made clear and presented to actual human experts so they can incorporate it in their diagnostic routines? I'm sorry to be cynical, but all I've seen so far from "medical AI" is spurious correlational relationships in data, sometimes even in noise, presented as a causal deterministic predictions. If you found a new way of predicting cardiac arrest, that's great. Let us know how.