If you look at almost any medical study, plots of the raw data of biomarkers of individuals are for the most part scattered all over the graph almost randomly, even in healthy individuals. Then the researchers typically draw some average graph somewhere roughly through the middle of all these random points and come up with predictions or sometimes prescriptions often based on whichever new pharmaceutical they are trying to create or increase a market for.
Being from the hard sciences, these graph interpolations are hilarious to me - you could often pretty much draw a graph of any shape you like by changing the assumptions used for interpolation. Grossly out of range biomarkers are one thing, but otherwise coming up with “optimal” recommendations in biology really is pretty much as much of a joke as herding cats.