Step 8: Record the data. That should actually say, “Record and act on the data.”
I’ve been recording all of my diabetes data—fingerstick BGs, CGM values, carb consumption, insulin delivery, exercise details, infusion set changes, etc.—for the better part of the last decade. I’ve been uploading, downloading, or otherwise saving it for the last few years. And thanks to Tidepool, Minimed Connect, and Garmin Connect most of this data is on the cloud, and a lot of it was put there automatically. I’ve also done the (hard) work to get all of the forms of data into the same place sync’ed up with each other. It’s great being able to see so many relevant things in one view.
What I haven’t been so good at is going back through and reviewing the data for patterns. I mean, I do know (generally) what’s happened in the past and have a vague intuition about what will happen if I change one thing or another. I could be much more systematic, though. In the past I’ve written down by hand a lot of the information that gets automatically stored for me (plus the all important “active insulin”/”insulin on board”/”IOB”) and then used that to figure out what might happen next. It’s a long and tedious task, but it’s worth it to jot down a few key details and see how results compare with each other.
Over the last couple weeks of basal testing, I have been manually recording values and looking for patterns. It’s not that hard, and I should extend it to exercise and the meals I routinely eat. Additionally, now that I’m trying harder to follow the rules or change them to be the right rules, I have the opportunity to gather higher quality data. (For a while a lot of the necessary context about where I was diverging from “the rules” weren’t adequately captured in my data, which makes it less useful.)
I’m still going to take the first cut at looking for the patterns and come up with new and better rules, but there’s also a role for software to aid with the analysis. I’m hesitant to lean on this too much, since I need the mindfulness of looking critically at my diabetes experience on a daily and weekly basis. Nevertheless, I’ve been exploring machine learning techniques recently, and that seems like an excellent reason to keep gathering the highest quality data possible.