UCSB Research Edges Closer to the Artificial Pancreas

Clinical study patients used an algorithm on tablet computers which maintained normal blood glucose levels in diabetic patients. In the future, the algorithm will be downloaded into a chip in the insulin pump itself, and no tablet will be needed.

Terra Hillyer cried with happiness when the continuous glucose monitor came on the market in 2006. It allowed her to see an automatic measurement of her blood sugar levels every five minutes, which would hugely improve her ability — she has type 1 diabetes — to maintain normal blood sugar levels and minimize the disease’s complications down the road. Today, the Goleta resident said she feels a similar landmark improvement in diabetes control is on the horizon — a big step toward an ultimate goal of diabetes research: the artificial pancreas.

Researchers at UCSB announced Wednesday that they have developed and successfully tested algorithms that will upgrade current insulin-pump technology so that pumps can provide automated, intelligent, personalized insulin therapy.

“Right now, most insulin pumps have programmable chips,” said Thomas Peyser, who is working with the UCSB researchers to translate the algorithms to the market. “We will simply be providing pump companies with the code that they can put into the chip to transform the current insulin pumps into an artificial pancreas device, or what we call an automated glucose control system.”

In a closed circuit, the algorithms — which would be personalized for each patient — will input information from the continuous glucose monitor to forecast the amount of insulin the body needs “12 moves into the future,” said Frank Doyle, chair of the UCSB Chemical Engineering department and lead researcher on the project. Doyle, who has been working on the algorithms for over 20 years, said it was as complicated as predicting moves in a chess game.

Frank Doyle (left) is the lead researcher from USCB, Fred Gluck (middle) is a retired CEO of McKinsey & Co, and Thomas Peyser (right) is the co-founder and chief scientist of Automated Glucose Control, LLC.

“The closed-circuit system provides much tighter control at an unprecedented level to minimize complications and to improve the quality of life,” said Doyle. “It will have immediate benefits, as it will lower health-care costs in the country and it will reduce the amount of decisions people with diabetes need to make on a constant basis.”

Hillyer said she sees the biggest hope in the automated system’s ability to remove the never-ending slew of choices she must make about the food she eats and the amount of insulin she injects.

“Down the line, this means there’s more hope in managing the disease,” said Hillyer, who participated in an early clinical trial at the William Sansum Diabetes Center — which worked with Doyle and his team at UCSB. “Diabetes is a 24/7 disease, and the closed circuit will take away all the decisions that I’m always having to make.”

The algorithms have had “very promising” success in the years of clinical trials conducted by Doyle’s team with the Mayo Clinic, University of Virginia, and academic institutions in Europe and Israel. While the average person with diabetes gets within the normal range of blood glucose levels only about 50 percent of the time, Peyser said, the clinical trials showed the algorithms raised that to over 90 percent at night and 70-80 percent during the day.

Doyle and Peyser presented their breakthrough technology at a well-attended MIT Enterprise Forum event Wednesday night at the Cabrillo Pavilion Arts Center.

“I think we’re within reach of the artificial pancreas,” Doyle said to the crowd of about 50 people. “We’ve figured out how to innovate model predictive control, we’ve tailored the model for patients, and we’ve provided the right safety envelope for our data.”

The automated glucose control systems could be on the market in about five years, Peyser said, but first the team needed to finish clinical testing and receive FDA approval. “Many clinical trials have shown its effectiveness, but we still worry about making it safe,” said Peyser. “It’s like flying a commercial airliner; you can’t afford to crash.”

Peyser doesn’t foresee any pushback from the large pump corporations already in the market, because, he said, they stand to benefit from the large increase in clients interested in the improved glucose control. Also, he said, the largest pump company in the United States, Medtronic, is currently developing an artificial pancreas system, with a prototype launched this week in Australia that is destined for the market by 2017.

“Currently, insulin pumps are only used by 10 to 20 percent of diabetes patients,” said Peyser, because “patients complain about discomfort and confusion.” But with the algorithms making decisions for customers and the pumps becoming easier to use correctly, Peyser said he believes “doctors will prescribe more pumps for their patients, and it will be used by a greater portion of the population.”

The first generation system will still receive patient input about meals during the day in order for the algorithm to accurately forecast future insulin requirements and will only operate fully automatically overnight. But later generations, Peyser said, would be closer to mature artificial pancreas technology and be fully automatic round-the-clock.

However, for Jim Sweas, a Santa Barbara businessman who has lived with type 1 diabetes for 50 years — against all his doctors’ predictions, the new predictive technology seems like something that cannot be tailored to fit each patient safely.

“How do you personalize something like this? I don’t think you can,” said Sweas, who attended the forum on Wednesday. “I’ve had diabetes for decades, and I know there is no pattern. I know that once you find your pattern, it changes as you age and with different activities.”

Still, Sweas said he would “venture a roll of the dice” and try the automated glucose control system if it came on the market.