I am looking for some advice on the most appropriate way to model some seed-based resting-state fMRI data at the group level. The data consists of three groups: Patients, Siblings, and Controls. I would like to be able to compare Patients and Controls, Patients and Siblings, and Siblings and Controls, with the idea being that then I can separate functional connectivity associated with disease vs associated with genetic vulnerability, and thus perhaps identify protective mechanisms in the Sibling group (whom obviously don’t suffer).
The Sibling group are related to the patient group, although the sizes are not even (i.e. for many, but not all patients, one of their siblings was scanned). Thus, the data are not balanced across groups, and there is some correlation between patients and their siblings (where they exist!).
Would it be possible (advisable?) to use 3dMEMA to model these data and include a covariate to account for a between sibling effect? Alternatively I have considered just doing T-tests to make the above comparisons.
For the comparison between Patients and Controls, and the contrast between Siblings and Controls, two-sample t-test should be good enough using, for example, 3dttest++.
The situation is a little tricky with the comparison between Patients and Siblings. One possibility is to try linear mixed-effects modeling through 3dLME:
Thanks very much for your advice. After a bit of playing around I’ve managed to get the model to estimate using the following script (see below). Is this the correct way to specify the random and fixed effects for what I am trying to do?
As I mentioned before not all patients have a sibling (and vice versa). Therefore, in the RFX part of the design matrix they have their own regressor (not shared with any other subject), does this mean I am essentially censoring them out of the model?
The first columns (“Subj” and “family”) are essentially the same thing. And 3dLME assumes that “Subj” is the only random-effects factor.
The
National Institute of Mental Health (NIMH) is part of the National Institutes of
Health (NIH), a component of the U.S. Department of Health and Human
Services.