‘Artificial pancreas’ uses algorithm to control diabetes in young children

Cambridge University researchers came to the conclusion after comparing the performance of the artificial pancreas, which uses an algorithm to determine the amount of insulin administered by a device worn by the child, against ‘sensor-augmented pump therapy’.

Management of type 1 diabetes is challenging in very young children because of a number of factors including the high variability in levels of insulin required and the way in which children respond to treatment differently, alongside unpredictable eating and activity patterns.

Children are particularly at risk of dangerously low blood sugar levels (hypoglycaemia) and high blood sugar levels (hyperglycaemia), which can affect brain development.

To manage children’s glucose levels, doctors increasingly turn to devices that continuously monitor glucose levels and deliver insulin via a pump, which administers insulin through a cannula inserted into the skin. These devices have proved successful to an extent in older children, but not in very young children.

Current technology – sensor-augmented pump therapy – requires parents to review their child’s glucose levels using a monitor and then manually adjust the amount of insulin administered by the pump.

But the researchers have developed an app dubbed CamAPS FX, which combines a glucose monitor and an insulin pump to act as an artificial pancreas. It automatically adjusts the amount of insulin it delivers based on predicted or real-time glucose levels.

It is a ‘hybrid closed-loop system’, meaning that the child’s carer will have to administer insulin at mealtimes, but at all other times the algorithm works by itself. There are no commercially available versions of fully closed-loop systems yet.

Professor Roman Hovorka, who worked on the project, said: “CamAPS FX makes predictions about what it thinks is likely to happen next based on past experience.

“It learns how much insulin the child needs per day and how this changes at different times of the day. It then uses this to adjust insulin levels to help achieve ideal blood sugar levels. Other than at mealtimes, it is fully automated, so parents do not need to continually monitor their child’s blood sugar levels.”

The researchers recruited 74 children with type 1 diabetes, aged one to seven years, to take part in their trial. On average, children spent around three-quarters of their day (71.6 per cent) in the target range for their glucose levels when using CamAPS FX – almost 9 percentage points higher compared to the control group that used sensor-augmented pump therapy, accounting for an additional 125 minutes per day in the target range.

Dr Julia Ware, the study’s first author, said: “Very young children are extremely vulnerable to changes in their blood sugar levels. High levels in particular can have potentially lasting consequences to their brain development. On top of that, diabetes is very challenging to manage in this age group, creating a huge burden for families.

“CamAPS FX led to improvements in several measures, including hyperglycaemia and average blood sugar levels, without increasing the risk of hypos. This is likely to have important benefits for those children who use it.”

In 2015, US researchers found that artificial pancreases controlled by software that run on mobile devices could face targeted security threats.