I didn’t expect to end up with #DIYPS when Scott and I first started building a system that would make louder CGM alarms. (If you are new to #DIYPS, read this post to learn more about it!)
But, as we added ‘snooze’ features for the loud alarms; the ability to input carbohydrates and insulin into the system; figured out that knowing your carbohydrate absorption rate is critical for managing diabetes as is the timing of the insulin activity in your body during mealtimes; and realizing that net IOB (factoring in temporary basal rates) matters…we ended up with today’s version of #DIYPS.
After the first 100 days, we knew #DIYPS was as good as the bionic pancreas was performing in clinical trials. And after several more months, I’ve had sustained results that show the value and impact of this type of system.
But, as more news and attention lately is focused on the “bionic pancreas” and APS (artificial pancreas systems), I have started to wonder… will those be good enough (for me)?
APS/bionic pancreas will not be a cure – it will not take away diabetes. It will make managing diabetes easier for many people.
But since I have #DIYPS, those systems may not be good enough (initially, for me).
The big difference between #DIYPS and APS is automation. With #DIYPS, I make all decisions and push all buttons to add or reduce insulin. There is still a “human in the loop”.
My hope would be able to add the ability for DIYPS to automatically set basal rates to zero temporarily overnight in advance of predicted low BG, to help prevent lows without having to wake up from a deep sleep and take that action myself. (And as Medtronic’s Low Glucose Suspend (LGS) technology has shown to the world and FDA, the risk of stopping insulin delivery for up to two hours while someone is sleeping is lower than it would be to continue providing insulin to them if they are not waking up and are low.)
If we’re able to make that happen with #DIYPS, I may be one of those people who doesn’t want to jump on board for day 1 of the first market available artificial pancreas system or bionic pancreas. (Although I reserve the right to change my mind :D). After all, at this point it would be a battle of the algorithms to see what system might work best for the most people. (And since I have my own algorithms that Scott and I developed, they might be better than the alternative for me.)
I also can imagine that there will be many people who won’t implicitly trust the/a system to make decisions for them all the time. They’ll still want a “human in the loop”.
#DIYPS-type systems that are “decision assist systems” may play an important role as a stepping stone platform for people to get comfortable with this new generation of technology that can predict lows and highs and make recommendations for them to take action.
Eventually, maybe everyone will be on an APS or bionic pancreas. Even setting aside the difficult questions of cost, availability, and insurance coverage, I know that one technology will not fit every person with diabetes (like how CGM in the Cloud is not for everyone); one system will not meet the needs of everyone; and none of us will be ultimately satisfied until we’ve prevented and cured all types of diabetes and eradicated the impact of any complications from diabetes.
But until we get there, #wearenotwaiting.
2 thoughts on “An artificial pancreas may not be good enough – #DIYPS perspective”
I share your concerns, Dana. I’m not sure I’m prepared to trust a machine to close the loop without improving the credibility of an open loop. Having the access to debug and verify the algorithms at work for you is easiest way this technology to earn the credibility needed to be used.