Single test: € = 1/p. Full warranty against multiple testing corrections only with a lifetime subscription.

Which sex does this person have?

Those tools may help us to find out why different people get different predictions (=the goal of descriptive modeling) not what are targets of interventions (=explanatory, causal modeling). Predictive models can probably be described (blackboxes only approximately), but not explained.

IMO, "explainable ML" is a misnomer. Should be "describable ML"

That's why every biostatistical method should undergo all four phases of methodological research before we can safely use it doi.org/10.1002/bimj...