If I’d got round to building the stuff I certainly wouldn’t want to move from Dash to O; you can control the algos and, if you want, you can factor in other things that control insulin sensitivity such as exercise (available on the iPhone through apple health, though maybe loop isn’t allowed to access it.) With O you are stuck with one algo.
I see O as an option for those of us who want to go closed loop but don’t want to hack to do it; I can certainly hack it, but there are other things I actually want to hack. And I am; I’ve got a bunch of stuff to hack my swimming pool and have been working on it for the last couple of weeks; this is something I actually want to do.
The Omnipod 5 starter box arrived and I’ve been wearing it for 2 days so far. Overall so far I rate it as an improvement over Dash but no verdict yet on the adequacy of the loop “automated mode”.
According to Insulet the algorithm adjusts itself over the first 2 or 3 pods. I plan to stay in Auto Mode as much as possible to give it a decent shot at self adjusting to my always surprising and sometimes baffling insulin needs. So I will wait until Pod #4 to start a new thread with some user review comments, after Mem Day weekend.
I think it’s learning the basal rate. The Medtronic closed loop system also takes a while to learn the basal. Tandem systems apparently don’t even try. The telling thing for me would be overnight drifts; for me they are a constant PITA because I end up waking up at least once every night to correct the drift.
My wife (non-D) has been wearing a CGM for the past 6 days and her closed loop system works really great - it invariably returns to 100mg/dL when she doesn’t eat for a few hours. This is what I can’t manage - over the same period if I get to 100mg/dL then I’ll go below that and, quite a lot of the time, end up having to eat to compensate for an impending low!
The advantage to the manual pancreas, however, is that my wife can’t bolus!!! Would you believe it; the designer didn’t include a bolus feature. As a result her blood sugar spikes to, typically, over 200mg/dL after eating, whereas my BG, managed by an intelligent use of the bolus, heads up to maybe 170 max (this is where her closed loop kicks in) and then drifts gently down.
I don’t think the closed loop feature is going to be valuable to people who aren’t fully developed T1s; you may well find it pointless, certainly that will be true for T1s who don’t need a basal at all. But it is going to be life saving for those of us who are fully developed and, at the same time, insulin pumps should be considered as generally applicable to the whole population - diabetic or not.
Yes that’s my understanding. The first pod starts using your user-entered basal rate but as time marches on will adjust. As explained to me, after “2 or 3” pods, your user-entered basals are no longer part of the auto-mode algorithm. So I will give it a chance to adjust its algorithm before reviewing the looping auto mode. But I will stay in auto mode as much as possible for those 2-3 pods to give the algorithm a good shot at learning/adjusting.
So far so good, the auto mode is truly “set it and forget it”. Occasional corrections needed along the way but nothing like manual corrections on Dash.
I don’t think that would work; the Dash interface allowed the bolus to be set manually, but I assume the pod also learns the BG correction factor, so overriding the bolus calculation is either going to turn off auto mode or mess up the correction factor. Giving a false carb value will definitely destroy any learning of the correction factor.
Yes the menu for bolus is pictured above. It is the same in auto mode and manual mode. It works fine if you leave the carbs and BG blank but I believe it adds some insulin on board to the auto mode.
Has anybody looked into sideloading the O5 app on unsupported devices yet? I know it worked great for the Dexcom app and I’d love to avoid having to buy and then carry around yet another phone.
Wow, lots of guessing and misinformation about iob, bolus, Basal and CF used in the algorithm here. Might be best to stop the guessing and wait a bit to really undersrand how the 5 will work and duscuss it with a knowledgeable professional…guessing and making uninformed assumptions us a rabbit hole.
Hi and welcome to FuD. There are some tech savvy people here who might have an answer. I have seen nothing about this anywhere. I think it’s too new for that app to be checked out by the tech crowd. I plan to wait for the app to support iOS for iPhone.
The Omnipod 5 PDM (which is free) so far seems less buggy than the Dash PDM. Battery life is decent, better than my iPhone.
The interviews with Insulet are out there for anybody to view. The professionals seem to be keeping most of the algorithm behind a curtain for us users to discover via trial and error. I got a similar response on my one call so far with tech support. I will continue the trial and error for a while and start a new thread with my observations.
With you in what you say. The published algorithms are, and have always been, very very very basic. Developing a real world algorithm to do this is, IMO, quite possible but it’s unlikely it will be documented in user manuals; a good starting point (sorry to quote wikipedia, search for Regression Analysis and don’t believe anything you read) is this:
The dependent variable is the amount of insulin to deliver.
Obviously Insulet and, for that matter [other companies], are doing this. Obviously they are not in a position to be able to explain it on Sixty Years. Alas they most likely aren’t going to make it FOSS (though they could, think about it); with FOSS we would just Laugh, Cry, Rant, Fix and, of course, Repeat.
i spoke to a rep from insulet a few weeks ago, they are working on apps that will work on ios and android 12, 13, etc, the app that was used in the clinical trials on the android was limited to mostly samsung phones that were on Android 11 and as long as they did not update to 12, it still worked, but sideloading probably won’t help because its the android upgrades that are causing the crashes…problem is the pod is linked to the pdm or the phone app, similar to looping, so if for some reason an update kills the app, the pod will be basal only/will need to be replaced…probably better off looping
