Now that we have Alma up and running with a DIY Loop system, all of her data live on a MongoDB database. Everything is in there – blood sugar (readings every 5 minutes!) and pump data (boluses, corrections, carb ratios, insulin sensitivities, basal settings, etc). This data could really help to figure out her settings – which would be AMAZING – but it’s going to take some time to figure out how to access it all.
But there are definitely patterns there, and I want to be able to see them!
So the chunk of code below is a Jupyter Notebook stored as a Gist on Github. Feel free to explore the repository in more detail. You can also check out this specific notebook on Github if you like!
In this notebook, I’m using a pretty simple custom python module (mdb_tools – you can find it in my repo) to access the database, load some collections, and then convert (some parts of) those collections to pandas dataframes to help me with analysis.