Dear Andrew

I was wondering whether it would be possible and sensible to use NBS to examine networks at the level of an individual participant? For example, I am interested in examining networks that are associated with clinical symptoms (yes/no responses to 40 items of a clinical rating scale) and resting-state EEG networks in individual children. I thought that I might be able to examine this by using the individual child's connectivity matrices from each resting-state epoch (e.g. 20 2-second epochs) and looking at whether there is a network across those trials that correlates with the child's clinical symptom scores. Though I'm not sure how to deal with the issue that there would be some children with fewer epochs than symptom item scores. Or perhaps this approach doesn't make sense at all?

Many thanks, Lizzie

Hi Lizzie, 

this sounds interesting, although it is difficult to provide advice without further background. If you have separate symptom measurements for each epoch, then this would be feasible and would involve computing a correlation matrix for each epoch. 

Alternatively, if you have multiple individuals in your study, you could test for an association between interindividual variation in symptoms and EEG connectivity. This is more typical. 

You might also want to consider establishing a normative model, which may be more appropriate if your goal is specifically individual level inference. 

Check out this review on normative modelling: https://pubmed.ncbi.nlm.nih.gov/31201374/

 Andrew
Originally posted by Lizzie Shephard:

Hi Andrew - thanks for your reply, the normative modelling looks helpful. I also found some work on Connectome-based prediction modelling, which seems to do what I would like to do - unfortunately I don't have separate symptom measurements for each EEG epoch, and I would like to try the analyses in the individual children rather than at the group level since I have quite a small group. Many thanks, Lizzie