Prove of efficacy in one patient?

[deleted_user]

I have a question regarding the validity of a statistical test. If we take N randomly sample from a laboratory sample of the same patient and we obtain N values of the same parameter before the treatment and then we repeat the same thing some time later, after the treatment, is it correct to perform an unpaired test between the N values before treatment and N values after the treatment, in one patient? Is this a valid way to prove efficacy, or the repeated measure of a parameter in one patient is rather a reliability measurement?

Thank you for your help.

[Doc_Duke]

see response under clinical trials.