Jothydev Kesavadev,MD
CEO & Director
Jothydev’s Diabetes Research Center
www.jothydev.net
jothydev@gmail.com

The dream of having an artificial pancreas is as old as the discovery of insulin. Artificial Pancreas is a misnomer. In the real world it is in fact a system of devices and mathematical algorithms wirelessly connected each other and with insulin infusions systems and Continuous Glucose Monitoring Systems. It was way back in 1964 that Dr. Arnold Kadish discovered the concept of insulin pump. Insulin pumps at that time was a huge device like a backpack delivering the hormone intravenously. But later on in the 1980’s clinical trials published by Dr. Pickup and Dr.Tamborlane  led on to the discovery of Continuous Subcutaneous Insulin Infusion. The early generation clinical trials for Artificial Pancreas were either with intra peritoneal or with intra venous route. Intra peritoneal delivery of insulin was more physiological which resulted in rapid action and dissemination of insulin. However such procedures required sophisticated surgeries often resulting in complications and aggregation of insulin in the infusion sets. Though intra peritoneal clinical trials went on for decades, now with the discovery of subcutaneous sensors there is a new path way which has opened the subcutaneous to subcutaneous Artificial Pancreas.

Minimed Medtronic was the first to prove the efficacy of subcutaneous bio sensors for Continuous Glucose Monitoring. The first one in this family was the CGMS Gold. And now we have both personal and professional Continuous Glucose Monitoring devices which are much smaller, accurate with no tubes and strings attached. The concept of Artificial Pancreas is to deliver the insulin from an insulin pump continuously and through the subcutaneous route throughout 24 hours in response to the blood glucose values which are obtained from the interstitial sensing via Continuous Glucose sensors. 

The major barriers to the discovery of Artificial Pancreas are with some of the major delays. One major delay is with rapidity of absorption of insulin. The present day's insulin though they are rapid acting are not adequate enough for subcutaneous delivery with insulin pumps. The second major delay is with action of the insulin which again is relatively slow via the subcutaneous route. The lag time which is seen with the present time Continuous Glucose Monitoring Systems is defined as the time lag between the interstitial glucose value and the real blood glucose value. This lag time will be much higher when the blood glucose levels are rapidly fluctuating. 

Until very recently the mathematical algorithms were run from either desktop computers or from laptop computers. The two commonly used algorithms in Artificial Pancreas are the Proportional Integral Derivative (PID) and Model PredictiveControl (MPC). However though the computers can wirelessly communicate with the Continuous Glucose Monitoring System and with the insulin pump, the patient cannot go around with such devices. For that we require a host to run the algorithms and the platform should be capable of hosting the sophisticated mathematical algorithms and easy to carry around. The platform should also be capable of wireless communication with devices and broad band connectivity for remote monitoring. All these requirements are met with the currently available smartphones. And hence the present day clinical trials with the Artificial Pancreas are with new generation smart phones which can display both the CGM and the graphs on insulin delivery. 
Though a completely closed loop system has not been commercially available, a partly closed loop which can shut down before the onset of a hypoglycaemia has already been in the market. This is the Medtronic Veo insulin pump with a low glucose suspect mechanism. 

The MPC algorithm in an Artificial Pancreas has got several merits. It has got the learning capabilities of learning the behaviour of the subject, the food habits, the timing and the historical pattern. During insulin delivery several glucose measurements are predicted before and several steps are planned ahead of time. MPC algorithm is often compared to that a chess game in which after the initial move the situation is again reassessed and the subsequent move will be planned based on the opponent move.

Artificial pancreas with artificial intelligence based on algorithms will be a blessing for type 1 diabetes subjects as well as those type 2 subjects with brittle diabetes. A truly artificial pancreas with a completely closed loop system still remains a dream. However the currently available smart insulin pumps and Continuous Glucose Monitoring Systems and devices where both are integrated together have already created a revolution in the difficult to treat diabetes.Advancement in technology is absolutely essential to resolve the present day challenges in the treatment of diabetes. However the extra cost of technology and the extra time to be spend with the patients will remain as the major challenges.