While I understand the issue in general it seems it should be relatively simple to validate that a closed-loop insulin pump is at least no worse. Isn't the obje

While I understand the issue in general it seems it should be relatively simple to validate that a closed-loop insulin pump is at least no worse. Isn't the objective of the closed-loop to just do finer control? So can't you just have the same hard limits as the normal pump and then allow the potentially buggy software to just fine-tune within a generally safe range?