Incorporating artificial intelligence (AI) into an artificial pancreas can enhance the system’s efficiency, new evidence has shown.
A first-of-its-kind study from the University of Virginia Center for Diabetes Technology has discovered that an artificial pancreas system which contains AI performs as well as an advanced experimental artificial pancreas system for keeping people’s blood sugar level in target range.
- Why night shifts work can raise risk of obesity and type 2 diabetes
- Ozempic butt: Women claim Ozempic caused their bum to ‘completely disappear’
- Thousands of harmful chemicals can leak into food from plastic containers
According to the developers, AI could be implemented in other devices with low computational power due to the increased efficiency it provides, such as an insulin pump.
Lead author Dr Boris Kovatchev said: “So far, this is the first clinical trial of a data-driven artificial pancreas system, which used an extensively trained neural network to deliver insulin automatically.”
An advanced experimental artificial pancreas system monitors and regulates blood sugar for people with type 1 diabetes automatically.
A total of 15 adults took part in the study by first using the advanced artificial pancreas for 20 hours and then the AI-supported artificial pancreas for the same amount of time.
The participants’ blood sugar levels stayed in target range 87% of the time when using the advanced artificial pancreas system and 86% of the time when using the AI-supported artificial pancreas, according to the findings.
In addition, the results have shown that the AI-supported artificial pancreas significantly enhanced efficiency and reduced computational demands by six-fold.
- Scientists develop experimental type 1 diabetes drug which shields insulin-making cells from immune system attack
- Changes in risk of arterial disease in people with type 1 and 2 diabetes
- Artificial pancreas: NHS rolls out hybrid closed loop systems for people with type 1 diabetes
The report states: “The AI-supported artificial pancreas is therefore more suitable for implementation in devices with low processing power, such as insulin pumps or pods.”
Dr Kovatchev added: “Neural-net implementation allows the algorithm to learn from the data of the person wearing the system. This opens the door to real-time, AI-driven personalised insulin delivery.”
The study was published in the journal Diabetes Technology & Therapeutics.
