Loop -- a dynamic answer to a dynamic problem

the Loop community has a chat forum that seems kind of hectic at times with several conversations weaving through the thread. Once you get the hang of reading it, however, it contains valuable info from people who care deeply about the topic.

Thank you, @Trying. Things are still not perfect for me. Last night I ate more ice cream than I should have. I was thinking, ā€œmaybe Loop can handle this.ā€ It was wishful thinking and reminded me that I still have diabetes! My overnight numbers were in the 180-200 range. I should know better to trust that overly adventurous side of my nature! Today is another day.

I thought about this idea when Medtronic announced news of 670 model with a fixed BG target of 120 mg/dL. If you told the system via calibration that a fingerstick 120 was really a 140 then, in theory, your idea should work. But that undermines safety when a 60 masquerades as an 80. I think your idea is worthy of a trial or two but if it doesnā€™t work, then youā€™re left with a four-year commitment to a pump you may not like.

Iā€™m not even going to ask, as I know it was worth it :grin:

2 Likes

Well, the only time I get a flat line like youā€™ve shown is when my sensor stops working! Boy, if only we could just have a few tastes of ice cream! What a luxury. I do that, too, just to test fate I guess, and always end up paying for it with high BGs. Good outlook - today is another day. I have to teach myself that, too!

@Terry4: I share your concern. It all depends on how Dexcom will implement the calibration. I believe that my first Dexcom system had a one point calibration. That means that it would assume a fixed gradient and would always work off the last calibration. With one point calibration it was best to calibrate in the middle of my target zone. I believe that Dexcom has changed to a multi-point calibration. That means that they take all recent calibrations into account and adjust the gradient dynamically. With multi-point calibration I expect my scheme to work. There are many ways to defeat my fudging. I will run experiments when I get the OmniPod/Dexcom AP. I will not switch to a tubed pump for AP.

Just wanted to add this graph to this topic. Iā€™ve now had good success with Loop three of the last four overnights. My one out-of-range overnight was the result of over-indulgence with ice cream. Anyway, last nightā€™s BG line was much like earlier ones with a flat steady trace. I didnā€™t have to eat any glucose tabs, a first for me on Loop.

yellow line=140 mg/dL, red line = 65 mg/dL

At this point, Iā€™m thoroughly impressed with Loopā€™s performance and I will continue using it. There are lots of new techniques possible with the flexible features of Loop. I would not have thought that varying the basal rate could have such a good effect on overall control. I will not bore you with more traces like this but will update if I learn anything interesting and useful.

Happy Thanksgiving!

3 Likes

Just curious to know how often the Loop needs to make basal changes overnight to get such excellent results? Iā€™ve started using HAPP, which implements OpenAPS algorithm on the Android. I use OmniPod so I use HAPP in its manual mode which means all temp basal have to be administered by me. It takes a huge effort to respond to all temp basal requests and I usually donā€™t respond to them overnight. This is just a first step for me with APS.

1 Like

Loop adds and subtracts to my programmed basal rates. Loop looks at the current status of various inputs every five minutes and then decides whether to stick with the current pump programmed basal rate or issue a low temp rate or a high temp rate. If you look a the Insulin Delivery graph above, you can see the changes that occur. I count about 58 basal rate changes from midnight to 7 a.m.

That would be tough to do during waking hours. If you tried to do it manually during the night, you wouldnā€™t get any sleep. The frequency of micro-adjustment to basal rates is better suited to a computer, not a human. That being said, making your changes in open loop fashion is a great way to get to know how the OpenAps algorithm works.

Most people use open loop operation for a time to see how appropriately the algorithm responds, given the settings such as insulin to carb ratio, insulin sensitivity factor, and duration of insulin action that youā€™ve customized for your metabolism. If these settings are not matched well to your actual needs then itā€™s better to detect this in open loop than it is when you close the loop.

Good luck with your experiments. These AP systems are amazing once you get them adjusted to your needs. Good luck!

All if this is so amazing, but it also makes me fear Iā€™ll eventually become completely out of touch with the best standard of care because I lack the technical knowledge and ability.

2 Likes

The current do it yourself pancreas systems require user learning equivalent to what a new insulin pump user needs to learn about starting fresh with an insulin pump and learning how to count carbs and adjusting your basal and bolus doses accordingly. Itā€™s a steep learning curve in the beginning but with time things become more routine. It will take me several months to be proficient with Loop. But it is a finite skill set that one can master given dedication and motivation.

The commercial systems, starting with the Medtronic 670 model will be a more turn-key arrangement for people who donā€™t want to get into the nitty-gritty of the current experimental systems.

This^^^^^^^^^!

And time.

So if it can keep you in range pretty much all the time and overnight, maybe except for mealsā€¦

How long will it take to build up the confidence in it to lose the cgm?

Canā€™t. The CGM is an essential part of the system. It canā€™t work without it.

Reading between the lines of your question, I think you equate additional technology like CGMs and pumps as burdensome. I donā€™t - at all. The most burdensome diabetes treatment I have ever used was syringe-delivered Regular and NPH together with the paltry feedback of four or five daily fingerstick BGs. This, to me, is not a simple system. You deliver insulin and somehow keep track of the timing and amount. You base your dosing decisions on the meager data of a few fingersticks. To me this is equivalent to flying blind.

Using instruments only to fly a plane without any visual cues is much better informed with data than using syringes, vials of insulin, and the occasional fingerstick. I know that Afrezza makes landings softer than liquid insulin but I trust no insulin well enough to deliberately under-monitor its performance, even if my control concern is primarily on the high BG side.

In my diabetes career, as technological complexity rises, the burden falls. I know that seems like an unreconcilable paradox to some, but to me itā€™s an obvious truth. @Dave26 gave the simple answer to your question. Removing the feedback portion (CGM) decouples the classic feedback control loop and an AP system cannot control anything beyond the direct human mediated changes.

1 Like

Thanks, @Terry4! This is so helpful. Yes, you are right, I could never mimic the closed loop APS with the manual, open version. I had no idea that it took that many adjustments but I guess there is a potential adjustment at each BG reading, every 5 minutes. My limited experience thus far with the open loop does indeed help me to evaluate my settings. So far Iā€™ve just been using my existing values and accepting the temp basal requests. I havenā€™t yet adjusted any of the settings. I hope you will keep us updated with your results, and most importantly, I hope you continue to have such excellent results, ice cream, aside of course:)

1 Like

What is the relationship of Loop to OpenAPS? Are they two totally different systems, or do you need them both.

Does Loop and/or OpenAPS work with the current version of the Veo? Thereā€™s a deal on right now with Medtronic in Canada where, if you buy a Veo, you get the new MM 630G for freeā€¦ Not sure if it would be a trade-in or what. Iā€™m in the market for a new pump, so want to explore all my optionsā€¦

These are two different paths toward the same goal. Loop uses Apple hardware and protocol. It was written primarily by one man, Nathan Racklyeft, and heā€™s since moved on to full time Apple employment and his Loop project is supported collboratively in an open source repository. OpenAPS is directly associated with Dana Lewis and her husband Scott Liebrand. The OpenAPS is designed around Raspberry Pi microcomputer. Dana has T1D and sheā€™s been looping for a few years now. OpenAPS is often set up with NightScout software, too. This is especially good for reporting.

I started pursuing the OpenAPS and got bogged down in motivational issues. The Loop is easier to replicate and itā€™s all wireless. Itā€™s quicker to go from starting the project to turning it on in a shorter time/effort span than OpenAPS. Relative novices have, with the help of both communities via chat, constructed their systems in a relatively short period of time.

What he said! V V V

1 Like

Both are DIY systems built on top of the same reverse-engineered (i.e. hacked) communications to Dexcom CGM and compatible MM pumps. Both require some work to setup.

They are different, and you do not need both. In short, the two setups are as follows:

  • OpenAPS = (code that runs on a small computer board*) + (pump communication device*) + (CGM) + (compatible pump) + ((highly recommended) Nightscout setup) + ((optional) watch)

*several different options are available for these components

  • Loop = (code that runs on iPhone) + (RileyLink**) + (CGM) + (compatible pump) + ((optional) Nightscout setup) + ((optional) watch)

**RileyLink is a small custom board, which can be purchased separately, and which works out of the box - no special skill required.

Most people use Dexcom CGM (G4+share, or G5), but the systems can also be put together using Medtronic Enlite CGM.

Here is a basic diagram that shows what one would need to setup Loop or OpenAPS:

You cannot buy a new compatible pump (with a possible exception of Dana R pump sold in Europe). An option available is to look for a used pump among the MM models shown in the diagram.

2 Likes

Thanks @Terry4 and @Dragan1. I think I much prefer Loop due to its accessibility. I tried setting up Nightscout this past summer and never finished due to its complexity and running into accessibility issues. I also have zero experience with hardware and, due to my visual impairment, am not confident Iā€™d be able to rig together customized hardware without some practice first.

I just did a search on Craigslist and there is a Medtronic pump listed, but it doesnā€™t say the model (although, judging from the picture, it looks like one that should work), cost, or location of the individual (other than they are located in my city). How do you obtain a secondhand Medtronic pump? Iā€™ll admit: Iā€™ve never bought anything off Craigslist and it makes me a little nervous not knowing where stuff comes from. Also, it looks like ā€œ554/754ā€ is the same as the pump marketed as the Veo in some countries, which confuses me a bit.