As I’ve written earlier than, I’ve noticed that once I’m not doing a number of exercise, my insulin sensitivity tends to lower.
While I’ve been managing that with automations that profile swap in keeping with location and time, I made a decision that it was time to combine stepcounts into Increase and see how which may change issues.
Making the most of work that Edward Robinson and Mathieu Tellier had undertaken, I’ve added step counts to Increase to experiment with how they is perhaps used.
Between Increase and AIMI variations of AAPS, we’re now gaining actual world expertise of the advantages and shortfalls of utilizing the assorted instruments within the Android ecosystem to make use of steps as an allegory for exercise and modify AID behaviour.
How are you utilizing steps?
Inside Increase there are three ways in which steps can be utilized.
Inactivity identification
Sleeping in
Exercise identification
Inactivity identification
Within the case of inactivity identification, as talked about earlier, when the stepcount over an hour is beneath a sure stage, a modifier is utilized to make insulin sensitivity decrease.
Proper now, every particular person must estimate the quantity that qualifies as inactive for them, however it gives a solution to begin figuring out when you find yourself being sedentary and make adjustments to the system.
Sleeping in
Inside Increase, the accelerated bolusing has a consumer outlined begin and finish time. By enabling the system to determine that you simply may not have woken up but, it permits the functioning of accelerated bolusing to be delayed.
This makes use of the beginning time entered by a consumer, and numerous steps beneath which the consumer is perhaps thought of to nonetheless be in mattress.
While that is helpful within the context of Increase, it could additionally present means by which period asleep and awake could possibly be recognized with restricted or no exterior enter and permit completely different behaviours presently.
I can envisage use to scale back foot-to-the-floor results, for instance, specifying that any steps above a sure quantity after a sure time set off a low temp goal, to scale back waking hyperglycaemia.
Exercise identification
That is probably the most tough of all of the settings to qualify utilizing stepcounts alone. While it permits the identification of a consumer strolling someplace, it doesn’t actually assist with another type of train, so is proscribed in its usefulness.
Integration with smartwatches might help, with the usage of different metrics, however these are sometimes not realtime into Android.
Proper now, Increase identifies numerous steps within the final 60 minutes, above which it considers exercise to be underway after which permits the consumer to outline a share lower that’s utilized to basal charge and sensitivity. There may be additionally the potential of integrating Increase disablement when exercise is detected.
Through the use of the final 60 minutes, the adjustments persist after the train has completed, when sensitivity is prone to be increased.
While this can be useful, it doesn’t have any help regarding present insulin on board, which performs a big half in train hypoglycemia, and it could be helpful to have the rise in step depend set off a reminder to eat one thing if steps are detected together with increased IOB.
It additionally doesn’t actually assist with actions that don’t considerably change steps, and use of issues just like the gyroscope, accelerometers and exterior coronary heart charge displays could also be helpful in that.
AIMI does one thing barely completely different with exercise, utilizing steps within the final 30 minutes because the set off, after which successfully decreasing profile by 60% and setting the next goal.
Does it work?
From earlier expertise, I do know that inactivity adaptation has made a distinction for me prior to now, however I’ve not been capable of take a look at it for lengthy sufficient to verify whether or not, on this kind, it does.
The sleep-in perform has labored very effectively, and I can see it being the precursor to alternative routes of dealing with day and night time in AID methods.
Exercise I’m much less certain about. I’ve very publicly said that I feel it’s extra advanced than steps, and I nonetheless assume that is true. As we begin to combine extra sensors, I feel it can change into extra useful, however proper now, for me particularly, IOB issues an excessive amount of for adjustment put up train begin to be vital.
Right here we’re proper on the sting of what we will do with AID methods. There’s little or no expertise of mechanically adapting for train and the way it adjustments what we do, however there are an enormous quantity of potentialities right here, and the training can be invaluable to each open supply and business methods.
What it does spotlight is the good thing about open supply. With open code, we will proceed to push the boundaries of what may be performed…