diverging color scale - way of treating lows if different from high BG levels
IDMVis
image from nightscout - horizon graph of average levels
XDrip+
feedback from the recording:
0:46 - Do yo uthink that hte treemap is useful?
Yes. overall, I really like this idea of separating out the patterns and click on patterns to see everythwere it occurs in all of the days, and getting a sense of the distribution of the different ranges within each of these patterns. A clinician might look at sharp falls or just falls in general, and then drilling down in to the ones taht result in extreme lows.
0:54
In the personal health informatics context of trying to help an individual with diabetes, modify their care or teach them, this is way too much time (the 300-ish days of data being rendered on the screen). The clinicians usually look at 2 weeks worth data.
1:01
I do see value in seeing extreme values. The green ones are not intertesting, these are not something that you would treat. But the bell curve that goes up and down is definitely something I'd want to pay attention to.
1:12
The bell curve can be used in more directly in the case of meals or other events. The reverse bell curve, I am not sure if it's long enough duration (150 mins), is useful for seeing the period of exercise. I think there is value in searching for patterns.
1:16
This is so different from what the clinicians are used to. The real power of this comes when you are trying to do a search across a lot of days. Yes I do see the possibility of this being valuable.
1:20 - 1:23
Removing the timelines in the treemap and only having them with a linear time x-axis. The bell curves do not have a shared x-axis. Non-monotonic patterns have a squiggle that is useful, but you can't really compare them.
Do you think it's useful to separate the monotonic and non-monotonic patterns?
Yes. The first thing I thought of was repeated treatments for a low where you jump up and fallback down, then maybe you start to over-treat it. I see value there.
