At present, the risk of hypoglycaemia presents a major challenge to tight glucose control in diabetes and most hypoglycaemia events happen at night, the researchers said.
The University of Cambridge team studied a closed-loop system in which continuous glucose measurement was used to adjust insulin dosing.
The system used an algorithm that was individually tailored and that adapted itself, correcting for errors in predictions and accounting for carbohydrate metabolism.
The researchers recruited 24 people with type-1 diabetes who were using insulin pump therapy. Participants were given a medium meal or a large meal with alcohol, to reflect realistic and testing conditions. Their blood glucose levels were then measured overnight.
Compared with standard therapy, those using the artificial pancreas spent 15% more time with blood glucose levels within target after the medium meal and 28% more time in target after the large meal.
Use of the artificial pancreas also cut by 3% the time spent in hypoglycaemia (defined as plasma glucose at or below 3.9mmol/L).
Reporting their findings in the BMJ, the researchers said: ‘An important feature of our closed loop algorithm is avoidance of hypoglycaemia, which if replicated in the home setting would revolutionise patient safety.’
They concluded: ‘The closed loop system has the potential to improve safety and efficacy of insulin delivery and may in future allow more flexible lifestyles in conjunction with improved glycaemic control for people with type-1 diabetes.’
An accompanying editorial said the use of a cross-over design, ‘the gold standard for clinical research’ mean the study ‘set a benchmark for future studies’.