From my experience, modeling and prediction methods are fairly standardized, involving the same maths across domains by leveraging linear regressions, etc. I see no reason why a model trained on these processes couldn't simply apply those same standard techniques. Math generally is universal after all.