Analyze data on Continuous Glucose Monitoring or Flash Glucose Monitoring in various Diabetes Technologies

[Maria Majeed]

Hello Everyone,

hope you all are safe and healthy

I have been trying to analyze CGM/FGM data for the past two months and have not discovered any way to do it in STATA. Although, I checked numerous scientific articles published in well-known journals, stating that STATA was used to analyze data on CGM/FGM.

is there anyone who has done it and can guide me step by step.?
Few initial issues are merging multiple patient files of extracted data from CGM/FGM devices and overcoming the repeated time error in the STATA.

I would be really thankful if someone can guide me or even if it's possible in any other software like R. I tried two packages in R but unfortunately, those are prone to repetitive errors and could not get through few errors.

Thanks

[Felix Bittmann]

Hey,
if the problem is extracting data from the devices then this is not a problem of Stata but of the device itself. Assuming that the data is already on your computer, what format is it stored in? You should aim for open formats, like .csv which can easily be imported. If this is a complex or technical process it might be best to contact the authors of the original papers you mentioned and ask them for advice since they have the expertise.

[Maria Majeed]

Hi Felix Bittmann Thanks

The problem does not lie in the format or extraction from devices. I have it in txt and CSV both.
it starts giving me errors when I used it even for a single patient after formatting the time variable, stating time has repeated measures (based on the fact that value is recorded every 15 minutes ) and data exists for more than 15 days.

[Felix Bittmann]

Could you please post all commands and outputs so we can see what is going on? I assume that this is longitudinal data so each patient has several measurement points. What analysis are you running? You could do a panel regression, then you need to set the time as so:

xtset idvar timevar

And then continue with the regression command xtreg.

[Joseph Coveney]

When you speak of "overcoming the repeated time error", do you mean that the sampling rate of the device is more frequent than one millisecond? If not, then there shouldn't be any repeated times, as Stata can resolve intervals of one millisecond and longer.

Also, what problems do you encounter with merging patient data files? This should be a fairly straightforward exercise.

Actually, I'm not sure why you would want to merge the device data dumps, themselves, inasmuch as you would typically condense the those into data summaries separately by patient—that is, extract features of interest in each patient's temporal profile (time series) of interstitial glucose level—and save the summaries patientwise. You'd then merge the latter. using patient (and date or time interval) as keys.