Related Experiment Video
Updated: Jun 9, 2026

Studying the Hypothalamic Insulin Signal to Peripheral Glucose Intolerance with a Continuous Drug Infusion System into the Mouse Brain
Published on: January 4, 2018
First in Human Feasibility Study of an Insulin Patch Pump Combined With CGM-Insulin Delivery Cannula
Hannah Cunningham1,2, Hanna Jones1,2, Steve Flint1
1Department of Medicine, University of Melbourne, Melbourne, VIC, Australia.
Aims:
Combining a patch pump (PP) with a single-insertion glucose-sensing cannula (continuous glucose monitor with an insulin set [CGM-IS]) may reduce user burden compared with tethered insulin pumps or multiple daily injections. We aimed to determine feasibility of a combined device.
Methods:
This feasibility study evaluated a PP with a CGM-IS (PP-CGM-IS) with manual bolus insulin dosing in T1D adults with insulin pump and CGM experience. The pilot phase involved 3 participants undergoing a mixed meal test (MMT) for 12 hours. In the second phase, 15 participants were studied for 72 hours, including day 1 and day 3 MMTs and one supervised free-living day at a hotel. Primary outcome was accuracy of the CGM device, assessed by mean absolute relative difference (MARD). Secondary outcomes included number of insulin delivery failures and device survival duration.
Results:
Twenty-six devices were inserted in 18 participants. Mean age was 51 years, 14 were female. Nine device (sensor) failures occurred immediately post-insertion requiring re-insertion. Mean absolute relative difference against Yellow Springs Instruments (YSI) in 17 devices was 11.6%. Consensus error grid analysis against YSI analyser showed 83.16% of readings fell within zone A and 100% within zones A and B. Mean glucose, time in range, and total daily insulin dose on day 2 of the main phase compared with baseline did not suggest compromise of insulin delivery. There were no serious adverse events.
Conclusion:
This first-in-human study confirms feasibility of integrating CGM-IS with PP. A PP-CGM-IS system warrants further research, including improved insertion process, greater sensor accuracy and ultimately real-time algorithm implementation.

