Cambridge University team release patient test results of insulin pump for artificial pancreas and monitoring device powered by AI.
Of all of the adjectives that could be used to describe the experience of living with Type 1 diabetes, ‘fun’ and ‘time-efficient’, would never make the list. The disease is congenital, with symptoms first manifesting in early childhood, and the process by which it is kept at bay requires manual monitoring of blood-sugar levels and regular administration of insulin.
For many with the disease this means taking a blood sample and (depending on the levels shown) injections of insulin. It requires constant vigilance – of diet, alcohol intake and exercise. Any steps beyond the narrow tightrope of viable glucose levels could plunge the sufferer into a life-threatening state.
What’s worse, the stresses (caused by the lack of automatic insulin regulation) accumulate over time, dramatically lowering the odds of patients living the long and healthy lives they could have otherwise had. The disease is a war – those who have battled it for long enough have even received medals [1].
Longevity.Technology: The artificial pancreas is part of a growing field trying to leap over the numerous obstacles and reach the golden ticket that will give patients peace of mind and enhance their quality of life. This new element of monitoring and AI leads this research a step closer than it’s predecessors.
The TRL score for this Longevity.Technology domain is currently set at: ‘Late proof of concept demonstrated in real life conditions.’
The TRL score for the technology addressed in this article is: ‘Technology has completed initial trials and demonstrates preliminary safety data.’
This is why the FDA’s 2016 approval of Medtronic’s MiniMed 670G caused such excitement [2]. The device was the first in a wave of new “artificial pancreases” that promised to monitor glucose levels and inject insulin into the bloodstream when needed. A huge advancement, for sure, and one certainly worth celebrating. But it was still missing one crucial feature: the MiniMed 670G still needed manual input from the user before it would inject.
So why not do one better and design an artificial pancreas which is “closed-loop”, automatically monitoring the patient’s blood and injecting insulin at the right time? This is exactly what has been done by Dr Roman Hovorka and colleagues from the University of Cambridge. Excitingly, it appears to have worked. Writing in the New England Journal of Medicine [3], Hovorka and his team outline their closed-loop ‘artificial pancreas’ as being comprised of an insulin pump and monitoring device powered by cutting edge AI. The AI makes predictions of future glucose levels to better time the automated insulin boost it delivers, allowing wearers to go about their day without the difficulties associated with manual monitoring.
In the trial of 136 patients conducted by Hovorka and his team, those assigned to use the prototype artificial pancreas spent an average of 24.2% more time with blood glucose levels in the healthy target range, as compared to those receiving manual injections. This could revolutionise care for those with complex-to-monitor cases of hyperglycaemia, like pregnant women and young children, before expanding out to become the go to method for treating diabetes [4].
All of this makes Hovorka’s tech a must-follow in the expanding bio-cybernetics market, with closed-loop artificial pancreases leading the pack at a forecast CAGR of 15.2% between 2018 and 2023[5]. Who Hovorka intends to licence his tech with still remains unknown, but his past collaborations with Abbott Diabetes Care, Medtronic, Dexcom and Animas make them all strong contenders for his final choice of partner [6].
“Our research opens a new era of glucose control in hospitals,” says Dr Hovorka. “Further research will establish the effect of improved glucose control on comorbidities, length of stay, and other clinically relevant endpoints, as well as whether certain patient groups benefit more than others.”
