Here is an update on how I use the devices in my daily routine, and how I plan to start data analysis once I finish recording:
Daily recording routine:
06:30-07:30: normal range of waking time. Remove/sync Oura and Wavelet and EEG electrodes (daytime data is too noisy to be useful, and it's pretty uncomfortable to wear the two leads on my forehead). Read glucose data (has to be done every 8 hours at least).
07:00-08:00: Start 45-60 minutes of moderate intensity running or cycling. Running can't be done with the gutrig on, so I'm trying to mix in days of riding a stationary bike. This will likely contain artifacts, but I'm attempting to minimize those (by remaining upright). Then remove gutrig, shower and replace. Sync data from the 24 hours before.
08:00-09:00: Small breakfast and or green tea. Takes about 20 minutes to eat/drink. Then to work/school (walk a couple miles, drive a couple miles)
11:00-12:00: Brunch/Lunch. First real meal of the day. Takes about 40-60 minutes to eat.
17:00-18:00 I usually get very hungry and go home (walk a couple miles, drive a couple miles).
18:00-19:00 Approximate dinner hours. Some type of dessert (dark chocolate or fruit) often follows
20:00 almost always have stopped eating/drinking by now.
21:30-22:30 approximate time I go to sleep. Apply forehead electrodes for sleeping. Put on Oura and Wavelet. Read glucose before bed.
Note on kinds of foods eaten/pattern of eating: I eat mostly Moroccan/California style food (I.e., tagine stews and salads), lots of fruit and scrambled eggs. Not much on grains, dairy, beans or sugary stuff. I usually have most of my food at an early lunch and evening dinner.
Chronotype: Obviously, I tend to be a morning person. My MCTQ (munich chronotype questionnaire) redundantly confirms I'm a lark...
I mostly work in Matlab (2017a) and Excel 2016. I will try to do analyses that highlight the temporal relationships among the outputs. Due to the limits of my skillset, a lot of what I do so far relies on visual inspection of spectra generated from my outputs. I plan to use continuous and discrete wavelet transforms (base code and papers mostly by Dr. Tanya Leise), and wavelet coherence from the matlab signal processing toolbox. The first will estimate the strength of the daily (~24 hour) and ultradian (~1-4 hour) rhythms I expect to see in the data. The second will compare signals from the outputs pair-wise to assess a) the stability of any phase relationships between each pair [e.g., the signals always rise and fall together (in phase, 0 degrees out of phase), or rise and fall opposite one another (antiphase, pi out of phase)]. I am interested in making very simple network models from the interactions among these outputs as well. I'll use non-parametric statistics to quantify the significance of any cool patterns I see (e.g., a significant change in the ultradian coherence of glucose and heart rate by sleep vs. wake state ). I plan to aggregate the data on day 11 (right after my CGM runs out).
Note on Freestyle Libre comparison: My two cgms are reliably ~ 10 mg/dL different from one another, sometimes up to 20 mg/dL different. The shapes of the signals at a glance look pretty similar though. They're about an inch apart on my arm.
Note on Sleep Quality while wearing all this stuff: About 03:00 I've been waking up, since doing this set of observations, and I'm almost positive it's because the gutrig is sort of irritating to wear (itchy?). I've been testing different electrodes for comfort (ConMed red dot, ConMed Softrace, ConMed ClearTrace2), and the Softrace electrodes are by far the winner (they're designed for use on infants I think). They have to be held on by a headband or ace bandage or they will fall off after a few hours, however. Last night I managed to only wake up for a few minutes, so fingers crossed my data wont be confounded by poor sleep every night.
If anyone has suggestions on analyses or features to look for, let me know!
I'll probably cluster shorter posts a few days at a time until I finish, in the spirit of keeping this as a project log.