They claim that it works with closed/looped systems too.
https://itunes.apple.com/app/id980437930
I personally haven’t tried this to the full extent but looks to have potential.
3 Likes
I tried it and found the interface to be lacking, even though its predictions were helpful back when we were using just a pump. Though my guess is they’ve improved the interface too!
1 Like
I tried it a few months ago but it was a non starter due to no real time CGM data.
Even with the new connecting to Share it only imports, obviously, Bg readings. Now, if they could have a real time import from xDrip+ including food, insulin, exercise and notes, then it would really be worth investigating again.
2 Likes
I talked to the CEO a few months ago and was interested. I understand that, for @docslotnick the lack of real-time CGM data was a big deal, though.
I am in the initial stage of trying PredictBGL and will post a few comments once I’ve given it a chance.
A question popped into my mind though as I was ignoring it’s first alert. PredictBGL became very concerned that my pre bolus for lunch exceeded 15 minutes. I normally wait 20 to 30 minutes depending on BG level. That first impression of the app got me thinking…what do I hope to get out of an app like this? Is it too heavily loaded with conventional wisdom aimed at less experienced bolussers to be of use to me? Will I spend more time ignoring alerts than managing my BG?
My answer to myself was that I want an app that is reliable enough to shortcut the “math calculations on the fly” part of diabetes. Sometimes before a meal I get a little touch of brain overload with all the variables I am bouncing around to make an educated guess about bolus dose and timing. I am hoping that this app is flexible and user friendly enough to become reliable for actually predicting BG. Curious whether others have tried this app and have thoughts about it’s usefulness.
Also FYI the app says it uses realtime Dexcom data but only with a paid subscription. I plan to try that out.
4 Likes
I would be quite interested from a technical point of view how the PredictBGL gains access to the realtime Dexcom data.
1 Like
I do believe it connects to your Dexcom Share account.
I wrote an email to their support to confirm this and an awaiting an answer.
1 Like
Yes, I’ve connected PredictBGL to Dexcom readings via share and the real time CGM data is automatically pulled in when connected to the internet. So far I’ve found the app useful but kind of a nag. It needs lots of entries for each meal, bolus, snack, activity etc. and if one is not diligent about checking every box it will assume a default value.
I am curious to see if it gets better at predicting BG… so far it’s predictions have been pretty far off compared to my gut feel method based on CGM trend. I plan to tweak my settings until I see predictions that are closer to my expectation.
1 Like
@John58 I downloaded the app just to check it out, but have not yet purchased a subscription so I can get the Dexcom readings in there.
So far I do find it a bit kludgy on my phone (a big screen Note 8). And after entering my information, I just got a notification to " enjoy your 45g carb dinner!" Looks pretty invasive and I could imagine it taking over my life.
I would be interested in using it fully for a couple of months to see if I can tweak it’s prediction features to match my reality. My goal is to ultimately limit my glycemic variability and I think this app could be of some help.
Keep us posted on your progress with it!
1 Like
I use it and I do get real time dexcom data… or at least nightscout. Syncs well with mynetdiary and the dex for me. I get lazy about logging food but it helps me keep track of insulin on board.
2 Likes
I find that I’ve figure matched insulin action pretty well but all of the glycemic index options even with logging protein and fat are too fast. I guess I’m just a slow digester… I’m much better at logging insulin than food anyway.
1 Like
The initial results are in! The graph below is typical for the app and shows the realtime Dexcom CGM trace (black line) versus the predicted BG (yellow line). The predicted line shows up as a solid black line on the iPhone screen. So far the predictions are pretty far off. The yellow dots are BG values manually input to the app. You’ll note that the app “updated” its prediction when a BG was manually entered but did not continuously update its predictions based on CGM BG readings. Not sure why but will investigate. I also suspect that the app does not automatically vary my insulin sensitivity after or during exercise which leads me to believe the I:C ratios or action time needs to be tweaked. There may be tweaks the user can input in the settings to account for this but I have not had time to review all those settings yet.
The interface on the iPhone is pretty busy but the “active insulin” value is easy to spot with a glance and seems to be the most useful feature to me at this point. The app also provides a recommended pre-bolus time for carbs that seems to be pretty reasonable so far, based on current BG and IOB.
The graph shows a day of moderate morning and afternoon activity (1/14) with a heavy workout (ice hockey) at 7:30PM on 1/14, followed by carbs/protein/beer that caused the 10:30PM spike and a correction bolus at about 11PM. The most reliable part of the prediction seemed to be following this 11PM bolus during the period when my insulin sensitivity was probably at its peak.
3 Likes
I discovered that operator error (me) caused some of those wildly high predicted BG’s on evening of 1/14. Here is a comparison of CGM (black) versus predicted (yellow) for today without any known snafus. The error I made yesterday is one of the PITA aspects of the app. The log is fairly user friendly when logging real time. When making catch up entries for things that occurred in the past, or posting an entry ahead of time (for example a workout), the app defaults to whatever your BG value is at the instant you made the entry. This app can not be used easily by busy people, the data entry requires paying attention to a lot of details.
2 Likes
So you log everything in MyNetDiary, and it imports it automatically into the app, along with the dexcom data?
1 Like
It looks like PredictBGL will import Dex data from Dexcom Share, and will import food data from MyNetDiary via Apple Healthkit on an Apple device. All of this in real time.
2 Likes
Yeah. If I bother to log food which I’ve been horribly bad about lately since my ratios are changing so much right now as my honeymoon is ending. When it stabilizes more I’ll log everything for a bit and try to nail down new ratios instead of my current guess and then eat to fit style.
I’ve got insulin effect and insulin on board to match up pretty well by fiddling with the acting time and the how much insulin does your pancreas still make settings. All the food absorption curves in the app seem to be faster than reality for me but I haven’t customized those at this point. Exercise factor is in inconsistent because my training is inconsistent.
1 Like
I wonder if it could automatically inform you of new carb ratio suggestions or insulin sensitivity. Would be nice feature.
2 Likes
The app makes suggestions for revised I:carb ratio and other factors. I have been tweaking those and will see if that helps the predictions. So far I have not figured out how to slow down the predicted BG rise after meals. Mine has consistently risen slower than assumed by the app.
2 Likes
I’m having some trouble with PredictBGL… It requires a lot of screen time for each meal, snack, correction bolus, exercise, adjusting ratios, etc. and I might not be the right PWD for that. The Dexcom interface is much easier to keep up with logging carbs, exercise and bolus with a glance and a few taps.
Soldiering on but sitting on my behind tapping data into an app is wearing down my ambition to give it a 30 day trial run.
2 Likes
Yea, they really need to practice “less is more”. Although live dexcom syncing is nice, there just needs to be a way to sync the pump in real time as well so that it can just compile the data and give you predictions or recommendations.
1 Like