Artificial pancreas automatically manages diabetics' glucose levels

By Dave Muoio
Share

A novel artificial pancreas (AP) system with a smartphone backbone could automatically monitor and adjust patients' glucose levels with little to no hassle, according to a new clinical study published in the journal Diabetes Care.

Unlike previously examined AP systems, the newest iteration — which consists of an insulin pump, continuous glucose monitoring receiver, a Bluetooth-connected smartphone device — uses cloud-based algorithmic calculations to determine weekly adaptations to patients’ insulin delivery without requiring assistance from users or clinicians.

“A limitation of current AP systems is that patients must still input carbohydrate estimations for delivery of meal insulin boluses,” the researchers wrote in the study. “As a result, a number of recent AP studies have included a period of clinician-led optimization of open-loop insulin pump settings or continued clinician adjustments to these settings throughout their use … To help with this process, different degrees of automated adaptation are being developed. Yet how best to optimize adaptation to safely improve glucose control … remains to be determined.”

The researchers’ efforts to develop an adaptive algorithm are based on a previous strategy called “model-predictive control.” By targeting an acceptable glucose level range and building an algorithm that improves with repeated daily cycles, the system can handle basal control and variations due to meals without manual assistance.

To test the automated system, researchers enrolled 30 adult patients with type 1 diabetes in a multi-site clinical trial. Over the course of the 12-week study they observed patients’ hemoglobin A1c (HbA1c) levels, a key indicator of blood sugar for diabetics, along with other secondary data generated by the continuous glucose monitoring device. The results, they wrote, were promising.

“This is the longest test we have conducted of our algorithm in an outpatient setting, hence it is a demonstration of the remarkable robustness of the algorithm,” Frank Doyle, professor and dean at Harvard’s School of Engineering and Applied Sciences, told MobiHealthNews in an email. “It is also the first time we have tested a real-time algorithm with adaptation, showing the power of personalizing the solution to the individual.”

Along with significantly lower levels and a reduction in hypoglycemia, the researchers also noted that less than 10 percent of algorithmic insulin delivery recommendations were overridden through the course of the study. In addition, the greatest changes to delivery were made in the early weeks of the study, with lesser adjustments becoming necessary as the algorithm adapted. Despite some limitations in study design and population, the system “has the potential to deliver enormous benefits” and warrants continued study into adaptive AP systems, they concluded.