Why don't we already have an Artificial Pancreas?

[slip]

We have the technology, pumps & CGMs so why is it so difficult to put the two together?

 

[tim2000s]

We have the technology, pumps & CGMs so why is it so difficult to put the two together?

Liability is the simple answer. I've been looking at OpenAPS for a while (and am desperate to get my hands on an appropriate pump to build one, having failed miserably so far) and the nexus of it being Open Source is that you have to put the bits together so any liability is purely incurred by yourself.

The companies making pumps/cgms and looking at AP capabilities have spent a large amount of time making sure they are comfortable about how everything works and have the appropriate safeguards in place. They have also undertaken vast amounts of trials to ensure their models work and are safe. The last thing they want is to be sued by the family of a T1D due to the AP giving that person too much insulin and causing death or a severe enough hypo to cause brain damage.

The basics of an AP algorithm are fairly simple, but ensuring it operates safely is much more involved and covers a great deal more pieces. That's been the main limiter.

The second is simply cost. Both pumps and CGM are not cheap, so in order to make the whole thing work, you have to invest a load of money and sell to a relatively small market. Recouping the costs of the above is extremely difficult.

 

[slip]

Cheers @tim2000s I knew you would reply.

The first company to get a functioning AP out would have great PR and kudos and the money would came after.....so long as they don't over charge for it!

I sort of understand the liability factor, but surely it's easy enough to program some safe guards in fairly quickly and cheaply. we have a pump/cgm system that will stop pumping when BG goes below a set level, do they cover the liability if you go in to DKA shortly after in that instance?

 

[tim2000s]

Cheers @tim2000s I knew you would reply.

The first company to get a functioning AP out would have great PR and kudos and the money would came after.....so long as they don't over charge for it!

I sort of understand the liability factor, but surely it's easy enough to program some safe guards in fairly quickly and cheaply. we have a pump/cgm system that will stop pumping when BG goes below a set level, do they cover the liability if you go in to DKA shortly after in that instance?

The first functioning commercial AP looks like it will be the 670G from Medtronic, which should get approval from the FDA in the US later this year. EMA approval will follow shortly after that. I believe that Medtronic are looking to get it to market in the US by the end fo this year, if all goes to plan.

The key issue with liability isn't the safeguards themselves, it's the what if the safeguards go wrong, so the manufacturers need to have 99.95% certainty that they won't, and really the only way to get any of this is by testing, a lot. We already have a pump system that turns itself off below a certain threshold and it works remarkably well. I've been using it for the last month or so.

Think about the following scenario:

A fat finger entry in the insulin sensitivity factor results in 1u for a 2.5mmol drop instead of 1u for a 5 mmol drop (this would be difficult with the Medtronic, but not out of the question). The AP then delivers too much insulin as the blood glucose level rises. The only way you have to stop it is the suspend function on this version of an AP. You have to ensure it drops in early enough to avoid a debilitating hypo.

The other potential issue is the lack of a 100% reliable glucose sensor. Using the 640G at the moment, due to the Smartguard feature, you are forced to calibrate the sensor every 12 hours. This is due to it being used for a specific medical purpose. This is likely to be the case on pretty much every AP out there, simply due to the critical dosing nature of the data.

The blocker to the AP that uses both insulin and glucagon (which would avert much of the above scenarios) is the lack of a stable and long-lived form of liquid Glucagon. Currently it is stored in powdered form, and the tests that have been done so far have required renewal of glucagon every day. That's why most of the plans for a dual chambered AP have quite a long lead time on them.

 

[slip]

Thank you for the very informative post Tim.

I guess just to throw in the mix is the QA insulin.................which we all know isn't that quick!

 

[tim2000s]

Thank you for the very informative post Tim.

I guess just to throw in the mix is the QA insulin.................which we all know isn't that quick!

To be honest, it's quick enough in terms of managing basal, although some of the new "quicker" insulins will be better for pumping. Ideally though, you'd have something with the performance of Afrezza, which would make it work far more effectively. Sadly, you need an intravenous or arterial link for that to be effective.

 

[brendan101]

Great responses Tim, on the money. #Afrezza!