The control group is needed to measure process properties / systematic error.

It does not need to be a placebo group. Previous data at the same hospital does not help because the regular medicine does not run according to study protocols.