We didn't see things calm down at all... less insulin is needed (no more 70+ unit per day) but variability was actually getting harder to control. When she started experimenting with Levemir she noticed that variability stabilized a bit, there were far fewer sudden high spikes. I could only guess as to the reason, whether it was pump sites being less consistent, or whatever. She is very interest in an AP system but none of the hardware configurations currently available are appealing. We want it to be with the dexcom sensor, and a wireless pump. Ideally, the Nightscout group will someday hack the pod protocols and design a new PDM that will interface to the Dexcom and not have any of the self-defeating restrictions that exist in the medtronic closed loop system.