Here is the next video from the Alternatives to Methodological Naturalism conference. In this video, Jonathan Bartlett describes how, at least in principle, one can model non-naturalistic causes mathematically.
Bartlett shows how modern mathematics yields functions which can be described but not predicted, and how they may relate to different modes of causation. He then uses randomness as a bridge case, and shows how randomness shows that such functions can be readily incorporated into models without harm to science in general.