This is correct - Closed loop control is not viable with the currently available sensor and pump technology. Current technology fails to meet all three of the non-negotiable requirements of stability of closed loop control systems, especially 99%+ reliability. Anyone who has seen what happens when a kid rolls over and sleeps on a sensor knows that "cutting off insulin when BG goes low" will likely result in DKA in cases like that. One could imagine that software could make up for these shortfalls, but it can not. It proves little that patients hospitalized for a 24 hour period had better BG control. Those patients were no doubt monitored by doctors or nurses 24 hours a day while on the closed loop system, because the sensors are not reliable enough to trust in a clinical setting under closed loop control. Perhaps for one or two days you could be lucky, but not for much longer than that. Even if the "Artificial Pancreas", which it is not, went on sale tomorrow, here's what life would look like for those who want to use it: First, you have to get a CGM and a pump. Second, you have to tell the pump when you eat, so it will compute boluses (due to insulin action time, meals can not be corrected after the fact under closed loop control - anyone who uses a CGM now and would wait to bolus until after the CGM detects the rise knows what the outcome would be). Third, you have to do BG tests every 8-12 hours and calibrate your "AP" Fourth, when BG goes low, carbs must be given (the "glucagon" approach talked about by the AP people is nothing but a concept at this point). Finally, you have to double check every decision the "AP" makes, because sensors are not reliable enough to be trusted under closed loop control. When BG drops overnight, you still need to roll your kid over and test their BG before deciding to "stop insulin delivery for 2 hours or until BG reaches a higher concentration". In other words, of the "Artificial Pancreas" went on sale tomorrow, nothing would change from what pump/CGM users already do today.