These are chat archives for nightscout/intend-to-bolus

3rd
Nov 2015
diabeticgonewild
@diabeticgonewild
Nov 03 2015 00:36
Well actually I can test it out, I think. I think the input values for those three variables are between 0 and 1. I just feel so terrible now. Drenched in sweat and heart is really palpitating still so going to take medicine to raise blood pressure. :/
So that is a bit of a relief though...those values being between 0 and 1 cause I can (possibly) avoid craziness on top of craziness...
diabeticgonewild
@diabeticgonewild
Nov 03 2015 01:33
Really want to test this out, but really feel crappy too. Going to try to test this out, even though I certainly messed up somewhere and it will be frustrating.
Jason Calabrese
@jasoncalabrese
Nov 03 2015 05:59
looks like this might work for me, put this cartridge in the 522 and seems like it would work with the inset 30 http://amzn.com/B00RR5YI3Q
Ben West
@bewest
Nov 03 2015 06:00
oh weird
diabeticgonewild
@diabeticgonewild
Nov 03 2015 12:39
If anyone is interested in ordinary differential equations and how they are solved for, check out lines 189-329. It's a very interesting way of programming things, so it's probably worth a look: https://github.com/diabeticgonewild/Artificial-Pancreas-2015-Mathematical-Model/blob/master/ODESystemSolve.m
diabeticgonewild
@diabeticgonewild
Nov 03 2015 19:51
@bewest A long time ago, you mentioned a possibility for #OpenAPS to possibly be integrated as an image with Cyanogenmod? What would be the advantage of that versus just having an app? Would it run, no matter what, as it would be set up to be embedded in the OS?
I am looking into options for versatility for #VirtualPancreas, which is an artificial pancreas. It will likely be implemented on a Raspberry Pi for prototyping first.
diabeticgonewild
@diabeticgonewild
Nov 03 2015 19:58
I am still on step 1 but I may be able to finish steps 1 and 2 before the new year. 1. Modeling (solving for parameters that allow a "diabetes patient" to be "copied" virtually for both simulations and and an artificial pancreas—given a controller—specifically a nonlinear model predictive controller—is programmed and used to represent the stochastic model—I have a copy of MATLAB code for a nonlinear model predictive controller but I have to modify it for a stochastic model and also want to add a Kalman filter or an extended Kalman filter to it for better BG prediction— http://etd.dtu.dk/thesis/248263/ep09_46_net.pdf ) 2. Programming and using the controller via vigorous and rigorous simulation tests, by using a "live feed" of Dexcom numbers to compare against OpenAPS decisions (I have a controller but it has to be modified for the stochastic model— http://etd.dtu.dk/thesis/248263/ep09_46_net.pdf ) 3. interfacing on the mobile scale and going live on myself
But probably going to bed soon. How pathetic of me. Getting tired and trying to listen to my body...FWIW, had a terrible day yesterday and trying to take care of myself...
Scott Leibrand
@scottleibrand
Nov 03 2015 20:16
Not pathetic at all. Take care.
Jason Calabrese
@jasoncalabrese
Nov 03 2015 23:11
think this got burried, any thoughts on if http://amzn.com/B00RR5YI3Q would work for me to use our animas inset 30s with the 522
think I might just buy it to find out
Tim Howard
@tghoward
Nov 03 2015 23:55
@jasoncalabrese the 1.8 ml reservoir is correct for the 522, but note that the way it latches to the pump is different than the Animas, so I think you'll need the pump end of the infusion set also.
That's how we switch between MM and Animas when we need to: we use the reservoir and infusion set tubing from the same manufacturer, but then the clip right at the infusion set seems to be universal, so we can swap manufacturers there.