I’ve finally found it. I have been trying to think of a good project to get my teeth stuck into – something that will get me up to speed with all these new Big Data technologies as well as being of some use – and I think i’ve got one: The Glucose Project.
In brief, the project aims to correlate blood glucose level and exercise as a way to help determine insulin requirements. But why on earth would I want to do that? As someone with type 1 diabetes, glucose levels are always on my mind. You have to constantly balance a number of factors (carbohydrates eaten, insulin, exercise, illness, stress) to keep your blood glucose level constant.
For me, one of the hardest factors to quantify is the cumulative effect of exercise on glucose levels over several days. During and immediately after exercise, carbohydrate and insulin requirements are reasonably predictable. However, in the days proceeding exercise my background (“basal”) insulin requirements may change by up to 50%.
My big problem is that during this period I have to adjust my insulin reactively rather than proactively, and my glucose levels might either be too high or too low, both of which are very detrimental to one’s heaIth.
What I plan to do is to try and correlate exercise, measured using a heart rate monitor, glucose levels (measured using a continuous glucose monitor), and insulin usage.
Now the amount of data I will be processing doesn’t in any way move into Big Data territory (see here for an interesting discussion on common misconceptions about Big Data and associated technologies). I could probably do it a lot quicker using some powershell scripts and Excel – but where’s the fun in that? That’s not really that important at the moment – this project will hopefully answer some questions about glucose management and allow me to experiment with some new technologies.
Next up – the plan.