help
help > RE: How to use "age" as covariates?
Jun 29, 2016 04:06 AM | Alfonso Nieto-Castanon - Boston University
RE: How to use "age" as covariates?
Hi Jeff,
If I am understanding correctly you have two subject groups (controls & patients), two sessions (baseline & follow-up), and one or several behavioral/cognitive variables measured both at baseline and follow-up (I am assuming here that these measures are obtained for all subjects, not just patients, please let me know otherwise), and you would like to know whether changes in cognitive/behavioral measures across time in patients might correlate with similar changes in intrinsic connectivity across time. If this is the case, then you could do that by selecting the 'patients' and 'patientsCognitiveChange' effects (the latter contains the change in cognitive impairment scores for patients between baseline and follow-up, and 0's for control subjects) and entering a [0 1] between-subjects contrast there, and then selecting the 'baseline' and 'follow-up' conditions and entering a [-1 1] between-conditions contrast there. That will show those areas where the changes in ICC scores across time correlate with changes in cognitive impairment scores across time in the patients group alone. If, in addition, you might want to know whether these associations are specific to the patients group (or whether, for example, some associations which may be present in the control group might perhaps not be present in the patients group), you would do that by selecting 'patients', 'controls', 'patientsCognitiveChange', and 'controlsCognitiveChange' effects and entering a [0 0 1 -1] contrast there (and again selecting both the 'baseline' and 'follow-up' conditions and entering a [-1 1] contrast). That will look at the same associations as before, now within both patients and controls separately, and show you those areas where the level of cognitive/connectivity associations are different between the two groups.
I am not entirely sure from your description whether you might also have multiple tasks/conditions (e.g. perhaps addressing different cognitive states) beyond the 'baseline' and 'follow-up' conditions mentioned above, so let me know if I may be missing something and/or any thoughts/comments.
Best
Alfonso
Originally posted by Jeff Browndyke:
If I am understanding correctly you have two subject groups (controls & patients), two sessions (baseline & follow-up), and one or several behavioral/cognitive variables measured both at baseline and follow-up (I am assuming here that these measures are obtained for all subjects, not just patients, please let me know otherwise), and you would like to know whether changes in cognitive/behavioral measures across time in patients might correlate with similar changes in intrinsic connectivity across time. If this is the case, then you could do that by selecting the 'patients' and 'patientsCognitiveChange' effects (the latter contains the change in cognitive impairment scores for patients between baseline and follow-up, and 0's for control subjects) and entering a [0 1] between-subjects contrast there, and then selecting the 'baseline' and 'follow-up' conditions and entering a [-1 1] between-conditions contrast there. That will show those areas where the changes in ICC scores across time correlate with changes in cognitive impairment scores across time in the patients group alone. If, in addition, you might want to know whether these associations are specific to the patients group (or whether, for example, some associations which may be present in the control group might perhaps not be present in the patients group), you would do that by selecting 'patients', 'controls', 'patientsCognitiveChange', and 'controlsCognitiveChange' effects and entering a [0 0 1 -1] contrast there (and again selecting both the 'baseline' and 'follow-up' conditions and entering a [-1 1] contrast). That will look at the same associations as before, now within both patients and controls separately, and show you those areas where the level of cognitive/connectivity associations are different between the two groups.
I am not entirely sure from your description whether you might also have multiple tasks/conditions (e.g. perhaps addressing different cognitive states) beyond the 'baseline' and 'follow-up' conditions mentioned above, so let me know if I may be missing something and/or any thoughts/comments.
Best
Alfonso
Originally posted by Jeff Browndyke:
Alfonso,
I've been following this thread because I would like to accomplish something similar but with an ICC analysis. How would one determine group-wise differences over time between controls and cognitive impairment level in patients? I'm interested in finding those areas that are positively associated with cognitive change over time and are associated with FC over time AND are significantly different from controls. My cognitive variable is continuous and mean-centered. From prior discussions I am contrasting our in-scanner tasks over time as (follow-up minus baseline), such that any resulting import values reflect decline or increase over time. I have three outside scanner behavioral/cognitive variables that are significantly different between groups from baseline, and ideally, I'd like to know how impairment/decline in the patients is related to areas engaged by our in-scanner tasks but also significantly different between groups.
Thanks,
Jeff
I've been following this thread because I would like to accomplish something similar but with an ICC analysis. How would one determine group-wise differences over time between controls and cognitive impairment level in patients? I'm interested in finding those areas that are positively associated with cognitive change over time and are associated with FC over time AND are significantly different from controls. My cognitive variable is continuous and mean-centered. From prior discussions I am contrasting our in-scanner tasks over time as (follow-up minus baseline), such that any resulting import values reflect decline or increase over time. I have three outside scanner behavioral/cognitive variables that are significantly different between groups from baseline, and ideally, I'd like to know how impairment/decline in the patients is related to areas engaged by our in-scanner tasks but also significantly different between groups.
Thanks,
Jeff