These are chat archives for nightscout/beta

13th
Oct 2014
Greg Waehner
@gregwaehner
Oct 13 2014 00:30
hey all - looking for any guidance on how to interpret the raw data from dex. I read Adrian's post of Tudiabetes, and am looking at my data. Just changed sensor and the "rssi" went from 190 and 948... filtered and unfiltered as well.... understand there's some algorithm to understand it? Anyone have pointers or reference I can read?
Greg Waehner
@gregwaehner
Oct 13 2014 00:37
And then the next two raw values were 181 and 168... but I know she's high right now.
Ben West
@bewest
Oct 13 2014 02:04
rssi should just be signal strength
should see it go up and down with distance from receiver, I would guess
Greg Waehner
@gregwaehner
Oct 13 2014 12:41
ok. thanks Ben. So then maybe I need to forget about RSSI... are 'unfiltered' and "filtered" the raw bg?
Ben West
@bewest
Oct 13 2014 18:41
think so... might be the raw "isig" eg the electrical current measured at sensor
is it a long numbr?
Greg Waehner
@gregwaehner
Oct 13 2014 22:50
Yes. Here's a snip:
Capture.PNG
Ross Naylor
@rnpenguin
Oct 13 2014 22:52
RSSI is a tricky little guy because it should go up as you get closer, but if you get too close or are next to another radio of some kind the value with change based on interference
Ben West
@bewest
Oct 13 2014 22:52
attenuation and harmonics
Greg Waehner
@gregwaehner
Oct 13 2014 23:09
right. But I'm trying to understand the "raw BG"... by process of elimination, it's the "filtered" and "unfiltered".... but does anyone know how to interpret the data values?
Ross Naylor
@rnpenguin
Oct 13 2014 23:10
you need to use the Calibration data to convert the filtered and unfilterd to the actual SG value
then you can use the slope and intercept data from that record y=mx+b to solver for sg
Greg Waehner
@gregwaehner
Oct 13 2014 23:12
ok. so for a "just changed" sensor during the start-up period, can I use the previous calibration data? Or do I need to use the calibration data from 7 days ago to match the "startup" sensitivity, etc.?
Ross Naylor
@rnpenguin
Oct 13 2014 23:13
the most recent cal data will correspond to the most recent record, then for unfiltered for example:
$scale*($unfiltered-$intercept)/$slope
= egv
Greg Waehner
@gregwaehner
Oct 13 2014 23:14
cool. thanks!
Ross Naylor
@rnpenguin
Oct 13 2014 23:25
:)