Debiotech, Swiss Academics To Develop "More Intelligent" Artificial Pancreas
Debiotech has signed an exclusive partnership with two Swiss research facilities to collaborate on a next-generation approach to artificial pancreas technology. The new system will pair a newly developed control algorithm with a micro-electro-mechanical systems (MEMS) pump, and the research team plans to improve infusion rate accuracy by anticipating patient needs based on their food intake, activity levels, and the time of day.
Artificial pancreas technology removes a lot of the guess work and effort currently required to manage diabetes, which involves regular finger-stick glucose measurement and insulin injections throughout the day. Next-generation technology — aimed at making these calculations and doses more automatic — could offer significant improvements to a diabetic patient’s quality of life.
Current systems for diabetes management include a continuous glucose monitoring (CGM) system that keeps tabs on glucose levels throughout the day, and an insulin pump that releases or stops insulin dosing based on those measurements. In June, Medtronic introduced the MiniMed 640G, which they believe makes significant strides towards a completely automatic artificial pancreas, compared to previous models.
Earlier this year, Debiotech acquired exclusive rights from iSense for the use of their CGM systems, and Debiotech CEO Frédéric Neftel stated that a closed-loop system has always been a part of Debiotech’s long-term strategy.
Of Debiotech’s most recent partnership, with Bern University Hospital and ARTORG Center for Biomedical Engineering Research, Neftel remarked in a press release that ARTORG scientists take an unconventional approach to computer algorithms, and their research may pave the way for a “more intelligent artificial pancreas.”
Stavroula Mougiakakou, head of the Diabetes Technology Research Group at ARTORG, maintains that the mark of a successful artificial pancreas is its ability to correctly adapt to the individual patient’s needs and anticipate dips and spikes in glucose levels.
“Approaches taken so far do not resolve fundamental difficulties: the patient’s variability, uncertainties related to system disturbances (e.g., food intake and physical activity), and errors related to the used devices,” said Mougiakakou. “The proposed algorithm is easy to use [and] introduces the concept of real-time personalization based on reinforcement learning. The machine learning method is able to tackle inter- and intra-patient variability and can compensate for the effects of uncertain events.”
The algorithm will be paired with Debiotech’s JewelPUMP, a MEMS-integrated insulin pump system placed directly on the skin, like a patch. JewelPUMP’s developers claim the system is highly accurate and ideally suited for an artificial pancreas system.
“The scrutiny on pump accuracy has increased in recent years. In vitro and In vivo, the JewelPUMP has shown its ability to inject the programmed dose very accurately,” said Laurent-Dominque Piveteau, COO of Debiotech. “Combining the algorithm developed by ARTORG with Debiotech’s JewelPUMP has the potential to revolutionize the way we approach the Artificial Pancreas.”
A recent pilot study published in the American Journal of Transplantation combined mechanical artificial pancreas systems with transplanted islet cells, which produce insulin. In a small population of 14 patients, investigators found that the system performed better than traditional diabetic injection regimens.
Aaron Kowaski, chief mission officer and VP for research at the Juvenile Diabetes Research Foundation (JDRF), told NPR that getting rid of devices is the ultimate goal in diabetes research, but in the meantime, mechanical solutions are going a long way towards reducing patient burden.
“So that’s what we’re trying to do,” said Kowalski. “Make smarter pumps until we have more nature-made solutions.”