|Perl: the Markov chain saw|
Re: Data averages by time of dayby diskcrash (Hermit)
|on Feb 26, 2004 at 23:43 UTC||Need Help??|
There are at least three separate issues to consider:
1. That blood glucose level is a function of food intake and exercise and that you probably can't easily take readings that reflect all of the effects, in a timely way. So the flow of data is probably undersampled.
2. That an average value is only one of several, relatively easy, metrics of interest. You can have a 3,4 or 5 point rolling average (a new data point came in, drop the earliest one in the average and move the data window along). You might also track minimum, maximum or the median ("middle point") during a 24 hour day.
3. Look at the mission. You track glucose to determine if your treatment is right. Is the right amount of the right medication, combined with the right diet and exercise controlling your blood glucose level? Tracking daily maxima and morning fasting levels might tell you a lot about how well you are controlled. Imagine also a "glucose control index" that you might invent. For example, (morning fasting level / maximum during any point in the day) and plot or track this ratio to see how your body is responding to diet and meds. Talk to your doc about what they feel are the most important attributes about glucose tracking. Some meds are really "peaky" where others provide more gradual control. These are the kind of features that can tell you about the underlying mechanism.
Best of luck to you and keep up the fight.