Software?

I'm a technologist by trade (specifically, a network engineer).  That means I keep the little green lights blinking. :)

I do technology for fun, too, and diabetes has afforded me the opportunity to learn more about databases and graphing.

I've written a quick CGI script that runs on my server at home and stores my carb, glucose, and insulin values in a database.

Once the data is in, the fun begins.  The beauty of a database is that you can have it automatically calculate averages and other formulas over a period of time.  I've got an A1C predictor, for example, that looks at a 90-day window of glucose averages and predicts an A1C result based on that.

Right now it's all very rough.  Build one to throw away, they say. :)  I'm using this as a kind of sandbox to try different things and see what I think works and doesn't work.  The database schema (the data structures and definitions, plus all the stuff that lives inside the database, like formulas) has been through a number of revisions already, and I'm sure it will go through more.

My next step will be to generate some HTML reports for date ranges and stuff like that (exports for spreadsheets, like Excel and OpenOffice.org Calc).  The step after that will be to start rendering graphs on demand.

So here is a question:  what statistics are useful in tracking and predicting how your diabetes needs to be treated?  One obvious thing I'm not modeling in the current schema is the ability to track multiple patients, for example.