As a diabetic, I often see patterns in my glucose levels over time that are difficult to pinpoint with traditional patient management software. This experiment takes a look at some of my favorite restaurants and their resulting impact on my glucose.
I used the Moves app to track my location and activity data for about three years. Using the Moves API, I was able to query a massive list of places that I'd tracked. I sorted them by visit count, and sorted through this list for some of my most-frequented eateries.
After assembling a list of about 20 restaurants, I queried my glucose database for each restaurant visit, starting one hour before my arrival and spanning six hours after. I then further filtered the list of locations to those with at least two sets of continuous glucose data.
I wanted to keep this one simple, so I focused on two pieces of data for each restaurant. First, I wanted to stack all of my glucose lines together to give a shape to my glucose for all visits to a restauarant.
After plotting the basic shapes, I finessed the glucose results to include one data point before the start of the time range, and one after the end to create continuous lines.
To finish, I calculated the average length of time spent at each restaurant and plotted that below the restaurant name. This not only gives context for the type of visit, but gives a visual scale for the time span represented by the horizontal axis. Some restaurants are quick to-go orders, while others are typically longer stays.
I like to use these experiments to make proactive changes to my health based on my findings. One obvious change might be abstaining from barbecue and tacos, which amounts to culinary heresy :)
I experimented with finessing my insulin coverage for the more problematic cuisines like Mexican. Through these trials, I was able to determine that giving 65% of my insulin at the time of the meal, and 35% extended over the subsequent 2.5 hours normalized the type of gradual spikes you'll see above for restaurants like Caramba Cafe.