Diabetes Health Center

Progress Continues Toward Artificial Pancreas

New Studies

Steil's team studied 10 children aged 2-6 with type 1 diabetes. At night, a computer-controlled system kept their blood sugar in the target range for more than five hours, on average.

In contrast, the kids spent only about three hours in the recommended blood sugar level range when a conventional insulin pump was used.

With a conventional pump, parents have to decide how much insulin to give based on their child's blood sugar level and how much they think the child will eat at each meal and at other times throughout the day.

"With kids' unpredictable eating habits, it's a real guessing game," says study head Andrew Dauber, MD, also of Harvard. Dips in blood sugar at night, during sleep, are one of the greatest risks faced by children as well and adults with type 1 diabetes, he says. The episodes can cause seizures and even sudden death.

Because they fear blood sugar dipping, parents may miscalculate the dosage and give their kids too little insulin, he tells WebMD. That can result in abnormally high blood sugar levels -- which over time, leads to all the complications of diabetes, Dauber says.

In the new study, children with type 1 diabetes were hooked to a glucose monitor. The results were entered into a computer, and a program calculated how much insulin was needed. "We did it manually for safety reasons," Dauber says. But the next version of the system will be automated.

The kids were much less likely to have episodes of abnormally high blood sugar at night when they used the closed-loop device than when their parents pumped their insulin. "Every parent told us they had the best night of sleep in years [when the artificial pancreas was used] because they didn't have to worry about hypoglycemia," Dauber tells WebMD.

Another system, called a Hypoglycemia-Hyperglycemia Minimizer, is being developed by Animas Corp. in partnership with the Juvenile Diabetes Research Foundation. In a new study of 13 adults with type 1 diabetes, the system was able to predict rises and falls in blood glucose and correspondingly increase and decrease the insulin dose safely over a 24-hour period.

"The cornerstone of the system is an algorithm that predicts where patients' blood sugar levels are going to go in advance of when they get there and change insulin [dose] accordingly," Anhalt says.

The system is the first to predict, rather than react to, blood sugar levels, he tells WebMD. A larger study is being planned.

These findings were presented at a medical conference. They should be considered preliminary, as they have not yet undergone the "peer review" process, in which outside experts scrutinize the data prior to publication in a medical journal.