Dear Anirudh Nath,

There is a study developed by Schmidt et al (2013) which deal directly with your subject. In their study, the system was based on subcutaneous glucose sensing and subcutaneous insulin delivery. Every 5 min, a glucose value was transferred from a Dexcom SEVEN PLUS (Dexcom, San Diego, CA) continuous glucose monitor (CGM) via a USB cable to a laptop computer. Two programs were running on the laptop: MATLAB (MathWorks, Natick, MA), which was running the control algorithm, and LabVIEW (National Instruments, Austin, TX), which was running a user interface that they developed specifically for CL studies by their group. Every 15 min, the CL system gives an insulin dose suggestion. The study physician needs to approve (or overrule) the dose suggestion and manually administer the insulin via the insulin pump (Paradigm Veo, Medtronic, Northridge, CA). Insulin was given as microboluses with 0.025 U increments in addition to a small continuous basal insulin infusion of 0.025 U/h. The latter has been implemented for practical reasons, as the pump cannot administer boluses if the basal rate was shut off. They have presented the DiaCon approach to CL glucose control and reported results and experiences from the first clinical study of their CL system. One of the greatest challenges met was irregularities in CGM input, and CGM inaccuracy was the direct cause of a hypoglycemic event. They concluded to improve system robustness, further improvements in sensor technologies are needed. In addition, they observed that a stochastic model identified from patient data, a robust controller, and adequately determined patient-specific parameters were important attributes for good and safe CL glucose MPC control.

In this way, I sent to you this paper (link).

Sincerely yours.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3876369/

In addition I sent to you another paper which deal also with this subject.

Best regards.

@ Marcelo Guimarães Silva 

Thank you sir.

I am very happy for your kind help.