Hello there, I moved this part of the discussion here so other people can find it more easily! The initial discussion is here.
Thanks for your responses so far, @evahamrud! What happens if I am actually interested on all 3 - the intervention, as well as whether the response differs by age and sex? This is why I was thinking I should check age and sex at baseline (time = 0) and post intervention (time = 1) first - are any age or sex differences in the metabolome at those timepoints? Then, construct an sPLSDA on the full data with a multilevel design.
I wished to do all these in one model for each dataset (baseline, post intervention, full), but from the links you sent me it doesn’t look like I can build a PLS model which includes all the variables of interest. Could you please expand a little bit more on how to look at the age and sex variables in my dataset, please? 
Many thanks in advance,
Evelyn