This article is more than 5 years old. Information may no longer be current.
Advanced algorithms, technology aim to improve on artificial pancreas systems
Click Here to Manage Email Alerts
AUSTIN, Texas — Hybrid closed-loop control systems are poised to revolutionize diabetes management, and new, advanced algorithms and technology aim to further improve on the latest artificial pancreas technology, according to a speaker here.
Regardless of age, most patients with diabetes are not within recommended HbA1c targets, according to Moshe Phillip, MD, chairman and chief scientific officer of DreaMed Diabetes, and director of the Institute for Endocrinology and Diabetes at Schneider Children’s Medical Center of Israel. In addition, hypoglycemia remains a problem for many patients, regardless of the tools used.
“Taking all of this into consideration, together with the burden of caring for patients with diabetes, it’s clear that we need something else,” Phillip said during his presentation at the American Association of Clinical Endocrinologists’ Scientific and Clinical Congress. “By many, it’s called an artificial pancreas, but actually, it is an automatic or semiautomatic insulin delivery system, based on a sensor, a pump and a brain.”
The “brain,” Phillip said, is an algorithm capable of predicting how insulin should be delivered based on past behavior. Currently, there are two main engineered algorithms used: the model of predictive control, or MPC, and proportional-integralderivative control, or PID, Phillip said.
But Phillip was looking for a more intuitive algorithm, he said: one that could play the role of endocrinologist and further refine insulin dosing based on a patient’s changing needs.
“When we came to the game, I brought in the engineers that I hired and told them, ‘Look guys, we are following 1,700 type 1 diabetics in our clinic, their median HbA1c is 7.6%, which means most are in nice control,” Phillip said, referring to the team at DreaMed. “And I anticipate that those that are in good control listen to me ... if you can translate that into an algorithm, then we can get everyone under control.”
“We started talking to engineers, and we couldn’t find common ground: What is a ‘normal’ [blood glucose] range?” Phillip said. “They asked, ‘What about 120 mg/dL, or 130 or 140 mg/dL, or 150 mg/dL? They needed a cutoff. Then one of them said, ‘I think the physicians are thinking in a fuzzy way.’”
From there, the “fuzzy logic” algorithm was born, Phillip said. Later renamed MD-logic, the algorithm is designed to emulate the way endocrinologists evaluate their patients, Phillip said, progressively refining how a patient responds to insulin treatment adjustments. MD-Logic, Phillip said, features an enhanced bolus approach, including control-to-range and control-to-target, and interplay of basal and bolus insulin with continuing corrections. Corrections are event-driven, he said, and learning and adaptation are personalized.
Glucositter, DreamMed’s artificial pancreas technology using MD-logic, became the first company to receive a CE mark for an artificial pancreas system in 2015.
“One of the things that I’m extremely proud of is the ability to learn and adapt,” Phillip said of the algorithm. “After every day, the system goes over the last day, mainly the insulin sensitivity, and learns for the next day.”
Phillip and his team wanted a system that could dose basal and bolus insulin, he said, because hyperglycemia is challenging.
“MD-Logic enhances the bolus approach,” Phillip said. “You still need to give a bolus before meals, but, if you forget, or decide not to give a bolus ... the system will bolus for you regardless of whether you have given it, because it is an event-driven system.”
In April 2015, DreaMed announced a collaboration with Medtronic to integrate MD-Logic and Medtronic’s PID algorithm together in a new advanced hybrid artificial pancreas system, currently under development. Under the terms of the agreement, DreaMed Diabetes will receive undisclosed royalties from future sales of each device utilizing MD-Logic. Medtronic will be responsible for all development and marketing of such devices. In addition, Medtronic has made a minority investment in DreaMed Diabetes of $2 million, according to the terms. A crossover study analyzing the PID algorithm vs. the PID plus MD-Logic algorithms is currently underway.
“Next is the combined algorithm running together, hoping to have an advanced hybrid closed-loop system hopefully toward the end of next year,” Phillip said. – by Regina Schaffer
Reference:
Phillip M. Moving Toward Artificial Pancreas. Presented at: AACE Annual Scientific and Clinical Congress; May 3-7, 2017; Austin, Texas.
Disclosures: Phillip reports receiving grants or research support, honoraria or consulting fees from Astra Zeneca, Bristol Myers Squibb, Eli Lilly, Kamada, Lexicon, Medtronic, Merck, Novo Nordisk, Pfizer, Roche and Sanofi, and holds stock in DreaMed Diabetes Ltd, NG Solutions and Nutriteen Professionals Ltd.