I've been thinking along these lines recently; we currently design treatment plans using statistical methods, i.e. averaging effects across 'representative populations'. But imagine if you took your car to the garage and they said "90% of cars showing this symptom respond well to this treatment".

When we truly understand how these biological systems work (which could be more than 100 years from now) we'll be able to measure what is different from the individual's baseline, rather than extrapolating based population averages.