Diabetes Week - Artificial Pancreas Project

It's Diabetes Week in the UK this week and after my rather forlorn post about the apparent disappearance of C8 medisensors non-invasive CGM last week I was pleased to read this little update on the Artificial Pancreas Project (APP) from Diabetes UK today.

It's still reasonably early days, and to my mind we are still quite a way off a 'real' management solution, but step by step people are cautiously being allowed to hand over the reins overnight to an automated system in their own home rather than under strict lab conditions with technicians standing by the bedside clutching Lucozade and NovoPens.

For those of you who have *no* idea what I am going on about, the Artificial Pancreas Project is seeking to 'close the loop' between two clever bits of diabetes kit. A Continuous Glucose Monitor (CGM) that can measure glucose levels and an insulin pump that can provide varying doses of insulin. In fact most people who happen to see my pump tend to assume that we are there already, and assume that a pump will just somehow 'know' how much insulin is needed and provide it automatically. Sadly it's not quite that easy, and while an insulin pump is a brilliant therapy option it does still require a fair amount of effort, guesswork and to be honest more than a little luck in balancing all of the hundreds of variables involved in day-to-day BG tomfoolery.

Part of the problem with the APP of course is that the technologies involved are improving, but still rather less than perfect. Continuous Glucose Monitors make an estimate of blood glucose levels based on glucose concentration in interstitial fluid (the stuff sloshing around between cells in tissue). This tends to lag a bit behind actual BG values, and those lucky enough to have access to CGM will know that what you are really getting is 'trend' information rather than the accuracy* of blood-based fingersticks. CGM manuals/marketing are usually pretty clear that any doses/corrections made should be done after confirming results with a fingerstick test. But the closed loop system needs to use this very same stream of trend information in order to make dosing decisions.

Currently the human trials of the APP are showing relatively promising results for looking after things overnight. To know that you had a smart system carefully tweaking your levels while you slumbered would be fantastic, but of course a full Artificial Pancreas would have to go much further and tackle the thorny issue of mealtimes. Personally I think we are a looooong way off that. For one thing current 'rapid' analogue insulins are hopelessly inadequate for the task. They take far too long to get going (1 hour for corrections to start taking effect anyone?) and act over far too long a time period to give a bunch of electronics, however smart, the ability to take over entirely.

What I would hope for instead is a commercially viable 'smart pump' which combines much of the effort and user knowledge/experience that currently goes into effective pump use, but with automated AP trickery to take the edges off the errors and irritations of daily pancreas impersonation. A system whereby you would bolus for meals, set TBRs for activity and generally do the basics, but the pump would be constantly monitoring how things were going and have the ability to tweak things and make corrections to keep things on a more even keel.

Now that really would be something.




* It should be noted, of course, that fingerstick BG tests in themselves are not particularly accurate and only really serve as a general guide as to what is going on.