It's probably reasonable to say that radiation actually being beneficial is very theoretical, but the linear no-threshold model of harm from radiation is definitely in the theoretical category too. It's used because it's very conservative, but evidence supporting it for low radiation dosages is very weak.