I went to Open Data Camp in the UK in 2023 and there was an interesting talk from two attendees with Type 1 diabetes about how locked down a lot of these devices are from a data perspective, which seems like a real shame - particularly when it's your own health data they're collecting!
CGM devices seems like the gold standard in terms of what they're natively capable of (and you don't need something separate on top of a flash sensor) and I think they're free on the NHS. Lots of them have companion smartphone apps, so you'd assume that it's possible to reverse engineer whatever API that these apps are using in order to get at the raw data. Dexcom seemingly does have an API, but it only works if you're located in the US.
From what I've gathered, the community has largely given up on trying to get at the raw data through Libre and Dexcom. That's not going to work in the long run, as Dexcom or Libre could cut it off at any time. The smarter solution is just to reverse-engineer the software component of these systems and take Dexcom/Libre out of the equation, or (even better) run a parallel system that does the data exporting you want. There are a number of open source solutions in this area. No guarantee that your physician will accept these, but they are pretty interesting, and very feature rich.
I worked for a company that had licenses for Libre, Insulet and Dexcom and we had their serial APIs - but you had to plug it in to your computer, store the data at our colo and use a Java applet to view it in the browser. In 2014. Unsurprisingly mobile apps have had more uptake since.
I talked to Matt about not owning our own data after GopherConSG where he gave this talk. It was enlightening how complicated the issue is - there's a lot of legal liabilities on the end of the data provider (the company that monitors the glucose) so I can understand why larger corps are a bit hesitant to open up.
On the other hand, it seems quite heinous that users don't have access to data that is rightfully theirs that they can action on it.
I never understood this. Would this be like buying a BIC pen and they have a required license agreement where any entropy made with the pen is owned not by the creator of the data but by the creator of the tool used? With out the person the data would not exists.
Perhaps if your bic pen fed your scratches into a system that ignored aberrant marks, cleaned up your letters, ran a spell check, and printed the result in the font of your choice?
This glucose device is a sensor. Sensors generally produce noise, sometimes a lot of noise, and require post processing in order to get the signal you wanted. The raw data is interesting but often not reliable as input.
That kind of processing is usually complex and companies often seek IP protection and close their source or use DRM. Whether that’s good or bad is a matter of opinion but the motivation is clear.
This is interesting, but is there a reason that you didn't explore the open source software in this space (Nightscout, xDrip, etc.)? These aren't new problems, and lots of people have solved them already with much more feature-full solutions that just work.
Many years ago, I went a similar route and built a small T1D monitoring stack from scratch for myself. It even pretended to be a Dexcom Follow client so I could get CGM from Dexcom in near real-time. When Dexcom eventually changed their internal APIs and broke my data ingestion, I decided to finally give Nightscout a shot. I've never looked back since.
As I see it, the big advantage of Nightscout is that it is a de facto standard interface, with many integrations already existing. And it's easy to build add-on apps on top of its API. I've built about four myself [0], [1] and there is a big community of users and developers building other things such as [2], [3], [4]...
Even though Nightscout is a little bit messy and requires MongoDB, it's surprisingly easy to self-host. I'm using the stock docker-compose file from the main repo with only minor modifications. I run it on a $6/mo VPS. As an alternative, there are two or three hosted Nightscout services costing little more than that.
I highly recommend you to consider going this "standard" Nightscout route because it can save you work (and worries) in the future, and you get to connect with the community around it. In my experience, going all alone from the start was not worth it.
After messing around with Elixir, I decided on my own that the Elixir ship wasn't worth it for me and I moved on to typescript.
Now I am debating between Rust and Go for backend development (for a web service). Which one between these two would be recommended for my needs?
My priorities are to be able to learn quickly and develop/iterate as fast as possible.
> […] wouldn't that decrease the speed of development for new people that come onboard?
If that was the case, wouldn’t people who love Go have created one after more than a decade in existence?
The reason why Go developers don’t like “frameworks” a-là Ruby on Rails in the Ruby ecosystem is because the Go community generally prefers libraries over frameworks because Go’s simplicity and flexibility allow developers to compose their solutions using small, composable packages rather than being constrained by a rigid framework.
This approach often results in more efficient and maintainable code. The Go philosophy emphasizes minimalism and encourages developers to avoid unnecessary abstractions.
Go is "amazing" . I am using Go since the very early days, and I absolutely love it. I have built services using Go and manage it all by myself. The community is so awesome and there is likeyly an opensource project that supports your need when you want it to do something exceptional.
I'd love to hear why Elixir wasn't for you - not from a judgemental place, I just like hearing others' experience reports for languages that stuck or didn't.
For a web service with fast iteration times, I'd probably pick Go of those two. While I'm not personally a giant fan, I think it is a much simpler stack that lets developers be pretty productive especially in the domain of web services.
Iteration was easier but the main complaint from my team was it was hard for junior engineers to pick up and understand. Along with that, IDE support still seems poor (we use Jetbrains and the Elixir plugin doesn't work reliably for us)
I just wanted to add that your writing style is wonderful and that this was a pleasure to read.
Incidentally, I work in pentesting, and one of my colleagues has Type 1 Diabetes; your overview of it and its resultant complexities really made me empathize with what challenges he has to surmount daily while still being one of the nicest, most approachable people in our team.
Your employer must be good then. Lots of companies don't hire and fire sick employees. Actually it's generally advised not to share one's medeical information at workplace.
I even read recently about the case of a worker who wasn't promoted despite everyone else expected him to be promoted. His sin ? He told briefly one of his colleagues that his car got stuck and he was tired of fixing it every now and then.
The promotion involves commuting for a certain distance. His manager told him he would have promoted him but he heard his car was not fit.
Actually, the sensor is super easy to hide (especially with long sleeves) and injecting is also quite discreet these days. I've seen injectors that are about the size of a marker or a wireless earbuds case. It's easy enough to go to the bathroom right before lunch or when you get an notification on your phone.
I'm not diabetic but I'm pretty sure what you describe would cause a lot of stress, especially if you don't want anybody in the office to find out... "hey you have a small bladder maybe?"
Stress has an impact on sugar levels, making it harder to regulate.
If you think knowing of the medical condition is bad for your career at your current employer, maybe it is better to switch employers.
In Germany, depending on the type and seriousness of your diabetes, you can get officially recognized as handicapped, and companies above a certain size (20 employees) are obligated to have a certain number of handicapped employees - so an employee who is handicapped, but can work (more or less) the same as a normal employee is actually a win-win situation.
Yes... you can lead a normal life with diabetes (you can even be a top level tennis player - https://en.wikipedia.org/wiki/Alexander_Zverev#Personal_life), but this article really made me appreciate the level of self-discipline it takes to do that.
There's also Loop on the iOS side: https://github.com/LoopKit/Loop. I've been using it for 7 or 8 years. There's also several pumps that have closed loop capabilities built-in, but I've found that they try so hard to play it safe that they are less capable than running your own.
Agreed. My daughter (5) is using a G6 + TSlim X2 closed loop. The target blood glucose is 6.0 mmol/L and you can't decrease it. However, I haven't tried (and likely won't try) anything DIY since it's my daughter and not myself.
Despite this, it is nice whilst my daughter is at school, since some automated corrections (even if overly conservative) are certainly better than none — teachers generally won't administer corrections (but will bolus for food).
As someone who’s planning to start a TSlim, this is a bit worrying. Admittedly even an average value of 7mmol would be an improvement over where I am, but I’ve certainly heard the lack of aggressiveness as an issue with the controlIQ system (if that’s what your daughter is using).
I do imagine it's different for everyone though, and would also depend on what your alternatives are. In our circumstances, it's a huge improvement over manual injections. But by no means the closed loop silver bullet we're all hoping for.
We ran the TSlim on BasalIQ (precursor to ControlIQ) for a couple of years. Haven't done her HbA1C since we swapped, so I don't have any hard data. Well, and with the constant changing lifestyle of kids, all data needs to be taken with a grain of salt. I think we're being woken by high alarms less often, but she just seems to be fairly high (just under 10 mmol/L) much more often. There are of course still run away highs much worse than that (don't want to misrepresent the situation!) and I guess they're somewhat mitigated by ControlIQ, but it definitely doesn't stop them outright, and manual corrections are still required if you want the high resolved in under 2 hours.
I kind of do have the option for a omnipod dash with a diy loop using something like nightscout, but my nurses have advised against it (I suppose significantly because it's not something they can directly monitor or have knowledge of). I'm highly tempted to just use the TSlim because if my projects are anything to go by it'll be years before I actually set it up haha
I'm in Australia so have had similar thoughts about the Dash. Because it's my daughter and not me, I figure I better not. When the Omnipod 5 lands in Australia we'll likely swap over since it supports closed loop out of the box.
I've recently switched from BasalIQ to ControlIQ on the X2. I very rarely feel like I'm on a "closed loop system". It's super conservative and left to itself it consistently keeps me higher than I'd like. With BasalIQ I could at least set up a low-enough target BG (5.0 mmol), now I'm stuck with 6.1 and the pump is actually happy to leave me at 7-9 mmol for hours, with very feeble interventions.
Long story short, I'm planning to buy a few boxes of Omnipod Dash pods, in a country where I can do so, out of my own pocket – so that I can try iAPS. #WeAreNotWaiting and all that :)
BTW Nightscout should not be required for Loop nor iAPS.
I suspect that type 1 has an environmental cause/trigger in many cases. As an anecdote, I knew many kids from military families or live in close proximity to specific bases where some chemical contamination may exist who have endocrine system issues without any family history of them. I knew very few from other areas and backgrounds with similar issues.
Edit: I'm curious if the people disagreeing with my musings have some studies investigating the topic, or some other interesting point.
Wow, ok, disagree with me even more without contributing to the conversation. I'll leave additional reading for you to disagree with this comment as well.
I don’t know why you’re being downvoted, environmental triggers eg from pollution are linked to autoimmune conditions. Any kind of excessive (oxidative) stress would increase the chances of it for that matter.
Incidentally, I created something related to this yesterday. A tmux plugin to display your glucose data as a status icon in your terminal.
As a T1 myself, I use a Dexcom as my CGM and have it working around that specifically, but I'd love to open it up to other devices too, and offer more. Feel free to check it out!
https://github.com/Cian911/tmux-xdrip
Sorry 1/2 into the article all that kept popping into my head is someone tell this man about a cgm like a dexcom . It all of this out of the box . My wife has been a t1d for 30 years the dexcom cgm makes all of this a lot better
He has that. The problem is not the CGM, it's everything around that (and the fact that both Freestyle Libre, and Dexcom, don't readily expose the data for the DIY types to do anything with).
While the Dexcom is "better" (than Freestyle Libre) in my opinion because the recent revision can talk directly to your phone continuously (the Libre only sends alerts, getting data requires you tapping phone to arm)... your insurance may cover only one or other. Either is way way better than nothing, but both are still a long way from automation.
Not sure exactly what you mean by "getting data" but recent versions of both Libre and Dexcom have apps where you can view graphs of data on your phone with no tapping step.
NHS likely doesn't pay for Dexcom. Someone has figured out how to make the Dexcom app broadcast values it's received, it works great: https://www.patreon.com/byod
It seems like you and the GP didn't read even the first 20% of the article. The author shows their CGM in the first 2 page lengths. They mentioned they did all of this because the official Libre app was constantly losing connection with their CGM. He bought a "hat" device called a "Miao Miao" and then siphoned off the data to his own CGM service that won't cut out if it receives, as the author puts it, "anomalous readings"
It has a side benefit of allowing them to show it on their Apple Watch using Google Calendar only.
Actually I did read the whole thing before posting... There are different CGM models where some require scanning and some report continuously. My reading was that the author's model seemed to only support scanning thus the need for the hat, rather than periodic updates to an app that newer models provide, but it doesn't seem 100% definitive from the article as I don't know if on newer models data export might require scanning? The point of my comment was that rather than all this work, getting a different device could more-or-less enable the desired functionality without so many workarounds and the bulky hat device by eliminating the whole scanning/inconsistency aspect. Though if you want to pull the raw datapoints etc. and export them elsewhere, then there are still more complications. The non-scanning ones might still be a bit unreliable as well, not sure where that aspect is coming from.
Have you looked into third party apps? I’m very personally used the app diabox (on android), I’ve heard suggah (if I spelt it correct) also supports a nightscout link like diabox. However I’m not sure how these work with dexcoms.
There's no cure, but 'remission' is possible in type 2 diabetes with lifestyle changes. Remission is defined as maintaining normal blood sugar levels without medication. This is not possible in type 1 diabetes
Type 1 here, there is a simple hack for both 1 & 2 and it's take melatonin at night. That upregulates the actual insulin receptors to do more with less insulin. I've been using that for 28 years now. My insulin requirements dropped 40% over my first 6 months and have stayed there ever since. I know two Type 2 who went into remission with diet, melatonin and who tapered off pills after initial use.
"There's no cure" for type 2 diabetes is true only on a technicality. Type 2 diabetes is defined as a progressive disease, which means that if you actually practically cure it, it is still technically defined as "in remission". It's just words.
Type 1 and 2 are quite different beasts - they both have in common that you pee sugar when it's not managed (diabetes means sweet urine in Greek).
Type 1 is when your immune system erroneously kills the cells producing insulin in your pancreas. Treatable by injecting insulin manually - but as any substitute, there are caveats, it's not perfect. Type 1 is pretty binary, you either have it or you don't, it appears rather rapidly, usually at young age, up to around 20. There's no way to avoid injections through diet.
Type 2 is when the cells that are _reacting_ to insulin are no longer doing so very well. Lots of reasons: genetics, body weight, physical activity. It's much less binary, with gradual onset over years. Can initially be managed by change of diet, activity, pills - later it also requires injections like type 1.
Type 2 is called Insulin Resistance - basically your pancreas works fine but the cells in your body stops responding to the hormone that tells it to remove glucose from your blood leading to a dangerous build up.
I use a drug called Metformin daily which slows down my digestive tract (you feel fuller after a meal longer like Ozempic), lowers the uptake of glucose in your gut and lowers glucose production in your liver.
Not essential. There are people who manage it or even get into remission with high carb diets (high carb low fat) too. One common thread across all these diets is avoiding processed foods.
I must note my gut has some interesting moments with it - some days - also I used to love mince (think US calls it a sloppy joe) - now I can't stand the sight and smell of it.
A family friend of ours just had a child -- 6 years old -- diagnosed with diabetes, and has to wear one of these monitors, with that same app and everything.
Scott Hanselman has blogged and youtubed about having type 1 diabetes over the years with various tech hacks. A while back he hacked it so his blood sugar showed up in his terminal prompt... https://www.youtube.com/watch?v=_meKUIm9NwA
Mildly related, I was diagnosed with obstructive sleep (OSA) apnea 6 months ago, and when sleeping I have to use a machine that blows air with a set pressure so it 'counters' the obstruction (a CPAP machine). Not mission critical like type 1 diabetes, but those machines log a lot more than I expected.
While reading this I wondered if someone has geeked something for similar for OSA, maybe for monitoring purposes or more severe cases.
I already know (and am incredible grateful for) OSCAR[1], but I'm really curious to see what else is out there.
Rather mission critical I reckon.
A CPAP machine was a life changer for me - long term untreated sleep apnea causes heart and other life issues.
During my sleep study I logged about at the worst 48 incidents per minute of stopping to breath (kind of difficult to do that when you tongue rolls back into your airway)
My sleep study revealed 90.7 events per hour (30 is considered severe) and 55% blood oxygen level. I've been using a CPAP machine for about 2 months now, but still feeling pretty tired in general. Still waiting for the life changing effects.
Have you analyzed the data from your cpap to see if it's doing what it needs to do?
It has the data for your apnea events, sleep time, all kinds of interesting things.
Check for leaks.
Make sure your mask is actually comfortable, I'd recommend trying multiple if you haven't already.
I was around 79 events and my CPAP was also life changing. But they’re not easy to get right.
First you need a mask that fits the profile of your face and the severity of your condition. There are a lot of different shapes and designs. Having tried a few, some were completely useless for me. Only one actually worked.
Facial hair also affects the performance of the mask too. I don’t know if this is something you need to be concerned about or not, but if you have a beard and wear a full mouth and nose mask, then you might want to consider a change of appearance.
Then you have the configuration of the CPAP machine itself:
+ how much pressure to apply?
+ what, if any, ramp up period to use before deep sleep pressure is applied?
Those two metrics took a bit of tuning before I found something I was comfortable with.
I was lucky that I had (and still have) an excellent doctor to guide me through all of this. Free from the UK NHS too. I can’t imagine having to navigate all of these caveats on my own.
Good luck with the CPAP machine. Not getting enough quality sleep can be debilitating. I’ve been in the same situation as you so understand how horrible life was before my CPAP machine.
I unlocked the machine by pressing two buttons and rebooting it and changed the ramping algorithm to less aggressive and increased the start pressure as it felt I was breathing from a straw when I tried to sleep.
I use a nasal mask (not the pillows) since I am not a mouth breather - for me the Fisher & Paykel Eson 2 (L) was just the right fit - took about two different masks before I settled on it.
I have recently be learning about tremendous success with well trained dogs alerting their owners as a complement to the technology based monitoring and alerting systems.
The article is very good and involves several things I'm passionate about including go and medical devices, I just wish software engineers realized how bad the Dalle3 stuff looks.
Yes! It's a great article, but why does anyone want a picture where the sun somehow manages to melt the glazing bar to illustrate an article about diabetes ? Or a big "Raise incident" button with all the other buttons being meaningless rubbish ?
Great article! While I haven't done anything as involved I have found using technology to track health issues to be empowering myself.
I'm curious about your thoughts on insulin pumps. I'm not very familiar but I know a few people with T1 that use them and seem happy with them. My understanding is that the idea is that they both monitor glucose levels and supply insulin to keep things stable. Have you considered them? Is there a limitation/downside to them that isn't obvious to a non diabetic? Thanks!
Pumps work alongside something like a libre and you’re correct they can deliver insulin.
Unfortunatley I’m not eligible for an insulin pump on the NHS so for now I’m sticking with injections. I do a good job of managing them this way so it works for me for now.
Pumps still require a fair amount of planning, experience, and knowledge about one's T1D in order to use effectively. The pump can only be reactive: sugar spikes, emit insulin. Sugar tanks, sound an alarm. You're supposed to tell it in advance how much sugar you're about to consume in order to keep things level, and they can't anticipate you deciding to exercise.
A good alternative to writing your own echo server and debugging requests one route at a time is requestbin, which will gladly take any requests you throw at it, log them, and optionally return a response of your choice.
Being familiar with PagerDuty and Datadog/ELK stack, all I want is a similar set of services that can scale (cheaply) to manage all people with T1D. Nightscout is a start but won't scale.
Thanks Graham. We actually spoke a little on Twitter after you posted that article here a year or so a go. I’m really happy to hear your son is doing well
Oh true, those days seem like a lifetime ago. I am now writing an app to make remote monitoring easier for Sams preschool, but are planning on writing a scalable replacement to Nightscout soon. Also digging into the various algorithms would be fun.
There are so many ways that software can make T1D more tolerable, I am glad you have found some solutions for yourself too :)
In you post you've mentioned that if toddler refuses to eat you need to prepare food. Did you try to give juice? When doctors explained to us what to expect, they said this would work just fine and kids love juice.
Yeah, but if you give juice or other yummy things every time your kid refuses to eat, then you are just rewarding them for refusing food. It is a very stressful balance.
Nice post !
I use a another set of tools to manage my type 1 and the one of my son.
Monitoring glucose:
Sensor libre 2 and xdrip mobile app using the bluetooth option where xdrip + oop2 reads values in 1min intervals.
( no miaomiao needed )
Values are pushed to nightscout
I got 2 nightscout instances: one for my self one for my son deployed to heroku.
Alarms are pushed to my phone for my sons low and high bloodsugar using pushover.
Heroku alarms are configured if service goes down.
Database: mongo atlas highly available.
We both use omnipod dash with android aps app.
A) helps with carb to insulin calc and carb counting, insuline on board.
B) closed loop, that auto corrects high glucose and stops insulin when too low
C) different profiles for things like "recently done sports" vs "didnt do much excerise in the last days"
D) aps app allows to track extra insuline injected by pen
bro, life saving stuff you've created there. also, for the "i need help" part, i'd try my damnedest so things won't go south like that (500 error). how about put it into a loop with backoff or something?
That's an amazing blogpost for techies!
Well written!
On the diet side, I'm living with almost no carbs for 2 years and over time my ability to turn fat into energy changed significantly, to the point I can do any activity I was doing before as long as I keep in check my sodium and potassium intake.
You can find people doing well with Type 1 and carnivore and you can find people telling you your insulin resistance will worsen.
I'd experiment regardless of what you find on the internet or the doctor tells you.
I spent two years of hell (not diabetes though, frequent diarrhea, stomach ulcers, blood in stool) following doctors indications and ended up with non sensical prescriptions and a low fat diet as a recommendation, when a high fat diet fixed all the symptoms in a week.
Camdiab (https://camdiab.com/faq) is available in some places (UK) to provide automated blood sugar level management.
I have dived into this a bit, and the thing that I grokked is the challenge of proving that the algorithm/device loops really work and are really safe. There has been a lot of painstaking work to develop and certify a thing like Camdiab. Unfortunately it also means that it's important to buy the right smartphone to make sure that the right libraries are available and being used by the app... Has been a bit expensive.
They have also walked the road of testing and proving the devices/algorithm for use in small children and teenagers. Bear in mind that this is all very complicated and there are a lot of differences in the behaviour of different hormone systems in different cohorts.
My son was on CamAPS for about 6 months with the Dana pump. Systems like that are even harder to get out because it relies on 3rd party hardware. Medtronic, Insulet's and a bunch of other systems are rapidly becoming a walled garden where the only way you can run the hardware and software you want is to run open source like AAPS or iAPS.
135 comments
[ 4.3 ms ] story [ 212 ms ] threadCGM devices seems like the gold standard in terms of what they're natively capable of (and you don't need something separate on top of a flash sensor) and I think they're free on the NHS. Lots of them have companion smartphone apps, so you'd assume that it's possible to reverse engineer whatever API that these apps are using in order to get at the raw data. Dexcom seemingly does have an API, but it only works if you're located in the US.
It's a patched app that broadcasts values it receives to any app that's listening, e.g. xdrip.
Libre 2 has also got patches floating around on the internet.
On the other hand, it seems quite heinous that users don't have access to data that is rightfully theirs that they can action on it.
Perhaps if your bic pen fed your scratches into a system that ignored aberrant marks, cleaned up your letters, ran a spell check, and printed the result in the font of your choice?
This glucose device is a sensor. Sensors generally produce noise, sometimes a lot of noise, and require post processing in order to get the signal you wanted. The raw data is interesting but often not reliable as input.
That kind of processing is usually complex and companies often seek IP protection and close their source or use DRM. Whether that’s good or bad is a matter of opinion but the motivation is clear.
As I see it, the big advantage of Nightscout is that it is a de facto standard interface, with many integrations already existing. And it's easy to build add-on apps on top of its API. I've built about four myself [0], [1] and there is a big community of users and developers building other things such as [2], [3], [4]...
Even though Nightscout is a little bit messy and requires MongoDB, it's surprisingly easy to self-host. I'm using the stock docker-compose file from the main repo with only minor modifications. I run it on a $6/mo VPS. As an alternative, there are two or three hosted Nightscout services costing little more than that.
I highly recommend you to consider going this "standard" Nightscout route because it can save you work (and worries) in the future, and you get to connect with the community around it. In my experience, going all alone from the start was not worth it.
[0] https://github.com/vitawasalreadytaken/koboscout [1] https://github.com/vitawasalreadytaken/glucoscape [2] https://github.com/kashamalasha/nightscout-widget-electron [3] https://github.com/mlukasek/M5_NightscoutMon [4] https://customtypeone.com/products/sugarpixel
Now I am debating between Rust and Go for backend development (for a web service). Which one between these two would be recommended for my needs? My priorities are to be able to learn quickly and develop/iterate as fast as possible.
If that was the case, wouldn’t people who love Go have created one after more than a decade in existence?
The reason why Go developers don’t like “frameworks” a-là Ruby on Rails in the Ruby ecosystem is because the Go community generally prefers libraries over frameworks because Go’s simplicity and flexibility allow developers to compose their solutions using small, composable packages rather than being constrained by a rigid framework.
This approach often results in more efficient and maintainable code. The Go philosophy emphasizes minimalism and encourages developers to avoid unnecessary abstractions.
https://github.com/mingrammer/go-web-framework-stars?tab=rea...
But you won't find anything that's anywhere near as comprehensive and batteries-included as Rails or Django.
For a web service with fast iteration times, I'd probably pick Go of those two. While I'm not personally a giant fan, I think it is a much simpler stack that lets developers be pretty productive especially in the domain of web services.
Incidentally, I work in pentesting, and one of my colleagues has Type 1 Diabetes; your overview of it and its resultant complexities really made me empathize with what challenges he has to surmount daily while still being one of the nicest, most approachable people in our team.
I even read recently about the case of a worker who wasn't promoted despite everyone else expected him to be promoted. His sin ? He told briefly one of his colleagues that his car got stuck and he was tired of fixing it every now and then.
The promotion involves commuting for a certain distance. His manager told him he would have promoted him but he heard his car was not fit.
and then there is penetration testing as in trying to break into things. is that what you do?
sigh
Despite this, it is nice whilst my daughter is at school, since some automated corrections (even if overly conservative) are certainly better than none — teachers generally won't administer corrections (but will bolus for food).
I do imagine it's different for everyone though, and would also depend on what your alternatives are. In our circumstances, it's a huge improvement over manual injections. But by no means the closed loop silver bullet we're all hoping for.
We ran the TSlim on BasalIQ (precursor to ControlIQ) for a couple of years. Haven't done her HbA1C since we swapped, so I don't have any hard data. Well, and with the constant changing lifestyle of kids, all data needs to be taken with a grain of salt. I think we're being woken by high alarms less often, but she just seems to be fairly high (just under 10 mmol/L) much more often. There are of course still run away highs much worse than that (don't want to misrepresent the situation!) and I guess they're somewhat mitigated by ControlIQ, but it definitely doesn't stop them outright, and manual corrections are still required if you want the high resolved in under 2 hours.
I kind of do have the option for a omnipod dash with a diy loop using something like nightscout, but my nurses have advised against it (I suppose significantly because it's not something they can directly monitor or have knowledge of). I'm highly tempted to just use the TSlim because if my projects are anything to go by it'll be years before I actually set it up haha
Long story short, I'm planning to buy a few boxes of Omnipod Dash pods, in a country where I can do so, out of my own pocket – so that I can try iAPS. #WeAreNotWaiting and all that :)
BTW Nightscout should not be required for Loop nor iAPS.
Edit: I'm curious if the people disagreeing with my musings have some studies investigating the topic, or some other interesting point.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8518604/
As a T1 myself, I use a Dexcom as my CGM and have it working around that specifically, but I'd love to open it up to other devices too, and offer more. Feel free to check it out! https://github.com/Cian911/tmux-xdrip
While the Dexcom is "better" (than Freestyle Libre) in my opinion because the recent revision can talk directly to your phone continuously (the Libre only sends alerts, getting data requires you tapping phone to arm)... your insurance may cover only one or other. Either is way way better than nothing, but both are still a long way from automation.
It has a side benefit of allowing them to show it on their Apple Watch using Google Calendar only.
I used to use Libre + MiaoMiao + Nightscout, but these days I much prefer Dexcom G6 + xDrip + Nightscout.
Pros for the G6:
- No 3rd party transmitter (e.g. the MiaoMiao)
- It's smaller and can go on the stomach so it's less likely to get ripped off
- 1 less app involved -- xDrip integrates directly to Nightscout and there is no need for the Dexcom app
- Supposedly the glucose readings are a bit more accurate than the Libre
Cons:
- Sensors are more expensive than the Libre
- The transmitter is not rechargeable so you need to keep buying them every 90 days
Type 1 is when your immune system erroneously kills the cells producing insulin in your pancreas. Treatable by injecting insulin manually - but as any substitute, there are caveats, it's not perfect. Type 1 is pretty binary, you either have it or you don't, it appears rather rapidly, usually at young age, up to around 20. There's no way to avoid injections through diet.
Type 2 is when the cells that are _reacting_ to insulin are no longer doing so very well. Lots of reasons: genetics, body weight, physical activity. It's much less binary, with gradual onset over years. Can initially be managed by change of diet, activity, pills - later it also requires injections like type 1.
I use a drug called Metformin daily which slows down my digestive tract (you feel fuller after a meal longer like Ozempic), lowers the uptake of glucose in your gut and lowers glucose production in your liver.
That and a low carb diet is essential.
Not essential. There are people who manage it or even get into remission with high carb diets (high carb low fat) too. One common thread across all these diets is avoiding processed foods.
- slow down stomach emptying (which was noted)
- improving glucose disposal
- reduce appetite by affecting brain chemistry
GLP1 RAs are really really interesting compounds (I'm a bit biased of course)
Neat idea.
While reading this I wondered if someone has geeked something for similar for OSA, maybe for monitoring purposes or more severe cases.
I already know (and am incredible grateful for) OSCAR[1], but I'm really curious to see what else is out there.
1: https://www.sleepfiles.com/OSCAR/
During my sleep study I logged about at the worst 48 incidents per minute of stopping to breath (kind of difficult to do that when you tongue rolls back into your airway)
Check for leaks. Make sure your mask is actually comfortable, I'd recommend trying multiple if you haven't already.
First you need a mask that fits the profile of your face and the severity of your condition. There are a lot of different shapes and designs. Having tried a few, some were completely useless for me. Only one actually worked.
Facial hair also affects the performance of the mask too. I don’t know if this is something you need to be concerned about or not, but if you have a beard and wear a full mouth and nose mask, then you might want to consider a change of appearance.
Then you have the configuration of the CPAP machine itself:
+ how much pressure to apply?
+ what, if any, ramp up period to use before deep sleep pressure is applied?
Those two metrics took a bit of tuning before I found something I was comfortable with.
I was lucky that I had (and still have) an excellent doctor to guide me through all of this. Free from the UK NHS too. I can’t imagine having to navigate all of these caveats on my own.
Good luck with the CPAP machine. Not getting enough quality sleep can be debilitating. I’ve been in the same situation as you so understand how horrible life was before my CPAP machine.
I use a nasal mask (not the pillows) since I am not a mouth breather - for me the Fisher & Paykel Eson 2 (L) was just the right fit - took about two different masks before I settled on it.
I have recently be learning about tremendous success with well trained dogs alerting their owners as a complement to the technology based monitoring and alerting systems.
I'm curious about your thoughts on insulin pumps. I'm not very familiar but I know a few people with T1 that use them and seem happy with them. My understanding is that the idea is that they both monitor glucose levels and supply insulin to keep things stable. Have you considered them? Is there a limitation/downside to them that isn't obvious to a non diabetic? Thanks!
Unfortunatley I’m not eligible for an insulin pump on the NHS so for now I’m sticking with injections. I do a good job of managing them this way so it works for me for now.
Lots of different implementations and hosts: https://duckduckgo.com/?q=requestbin
He is now on Dexcom+Omnipod+AAPS and doing well.
Being familiar with PagerDuty and Datadog/ELK stack, all I want is a similar set of services that can scale (cheaply) to manage all people with T1D. Nightscout is a start but won't scale.
Awesome work and best of luck, T1D is shite.
There are so many ways that software can make T1D more tolerable, I am glad you have found some solutions for yourself too :)
Monitoring glucose:
Sensor libre 2 and xdrip mobile app using the bluetooth option where xdrip + oop2 reads values in 1min intervals. ( no miaomiao needed ) Values are pushed to nightscout
I got 2 nightscout instances: one for my self one for my son deployed to heroku.
Alarms are pushed to my phone for my sons low and high bloodsugar using pushover.
Heroku alarms are configured if service goes down.
Database: mongo atlas highly available.
We both use omnipod dash with android aps app.
A) helps with carb to insulin calc and carb counting, insuline on board.
B) closed loop, that auto corrects high glucose and stops insulin when too low
C) different profiles for things like "recently done sports" vs "didnt do much excerise in the last days"
D) aps app allows to track extra insuline injected by pen
On the diet side, I'm living with almost no carbs for 2 years and over time my ability to turn fat into energy changed significantly, to the point I can do any activity I was doing before as long as I keep in check my sodium and potassium intake.
You can find people doing well with Type 1 and carnivore and you can find people telling you your insulin resistance will worsen.
I'd experiment regardless of what you find on the internet or the doctor tells you.
I spent two years of hell (not diabetes though, frequent diarrhea, stomach ulcers, blood in stool) following doctors indications and ended up with non sensical prescriptions and a low fat diet as a recommendation, when a high fat diet fixed all the symptoms in a week.
I have dived into this a bit, and the thing that I grokked is the challenge of proving that the algorithm/device loops really work and are really safe. There has been a lot of painstaking work to develop and certify a thing like Camdiab. Unfortunately it also means that it's important to buy the right smartphone to make sure that the right libraries are available and being used by the app... Has been a bit expensive.
They have also walked the road of testing and proving the devices/algorithm for use in small children and teenagers. Bear in mind that this is all very complicated and there are a lot of differences in the behaviour of different hormone systems in different cohorts.