A: Great question. As you may know, JDRF has established a consortium of top diabetes scientists, doctors, and engineers at 8 research institutions to develop and test sophisticated algorithms that can automatically control glucose levels in a variety of circumstances and types of people. This research builds on work already funded by JDRF at Yale University, which showed people with type 1 diabetes using closed-loop systems in the hospital spent more time in normal glucose range. The consortium is now funding multiple sites to test the safety and effectiveness of various versions of a closed-loop system. While the initial research will take place in hospital-based clinical settings, the goal of the initiative is to eventually test artificial pancreas systems in everyday settings such as home, work, or school.

There are two main types of algorithms that the JDRF Consortium members are studying. The first is called PID or Proportional-Integral-Derivative. This type of algorithm looks at a target glucose level -- say 100 -- and then the actual glucose level (the proportion away from the target), the rate of change of the glucose (the derivative) and the amount of time spent above or below the target (the integral) and uses this information to make automated insulin dosing decisions. While this is not customized to each individual, the glucose is being monitored continuously. The data we've seen thus far in JDRF supported human trials suggests that this works very well and that people on this type of closed-loop system do much better than when they manage their diabetes themselves.

Another sophisticated approach is called MPC or Model Predictive Control. This type of algorithm could actually learn about each individual patient and become "smarter" over time. The data that is gathered over time helps the algorithm better understand when the person with diabetes tends to get high or low and can become more or less aggressive during certain periods of time. Again, the data we've seen in the JDRF Artificial Pancreas Project Consortium using this approach is very strong and promising.

I think the one final key point to make here is that we don't need to be perfect to do much better right now. The average person with diabetes spends a lot of time each day with high blood sugar (over 9 hours a day if your A1c is 7.0%) and these control algorithms could help improve that while minimizing risk of getting low. I believe the first closed-loop artificial pancreas systems will use these sophisticated algorithms with help from the person with diabetes and this will allow us to do much better. The end result will be lower A1cs, far fewer low blood sugars, and less burden on the person with diabetes!