help
81 Subscribers 
help > Multilayer network statistics
Showing 1-2 of 2 posts
Oct 15, 2020 05:10 PM | Giuseppe Pontillo
Multilayer network statistics
Dear NBS experts,
I have a cohort of patients for which I obtained both functional and structural networks. My idea is to model the brain connectome as a multilayer network (as in https://pubmed.ncbi.nlm.nih.gov/28327916/) of two layers (SC and FC). Next, the objective would be to test for possible multilayer network differences between patients and a group of HC, as well as to find multilayer subnetworks which correlate with specific behavioral outcomes.
Is there a way to implement the NBS approach with multilayer networks? As far as I understand, it should be possible to use the NBS approach on a 2D representation of the multilayer network (the supra-adjacency matrix). Would it be appropriate?
Thank you,
Giuseppe
I have a cohort of patients for which I obtained both functional and structural networks. My idea is to model the brain connectome as a multilayer network (as in https://pubmed.ncbi.nlm.nih.gov/28327916/) of two layers (SC and FC). Next, the objective would be to test for possible multilayer network differences between patients and a group of HC, as well as to find multilayer subnetworks which correlate with specific behavioral outcomes.
Is there a way to implement the NBS approach with multilayer networks? As far as I understand, it should be possible to use the NBS approach on a 2D representation of the multilayer network (the supra-adjacency matrix). Would it be appropriate?
Thank you,
Giuseppe
Oct 15, 2020 10:10 PM | Andrew Zalesky
RE: Multilayer network statistics
Hi Giuseppe,
Linking structural and functional networks using a multilayer framework sounds like an interesting idea.
The NBS does not provide capacity to explicitly model multilayer networks. However, from the perspective of a connectivity matrix, multilayer networks are no different from any other network. So, it should be possible to run the NBS on the connectivity matrices from your multilayer networks.
But the concept of subnetwork can become challenging to interpret, given that it can now include a mix of structural and functional connections.
Andrew
Originally posted by Giuseppe Pontillo:
Linking structural and functional networks using a multilayer framework sounds like an interesting idea.
The NBS does not provide capacity to explicitly model multilayer networks. However, from the perspective of a connectivity matrix, multilayer networks are no different from any other network. So, it should be possible to run the NBS on the connectivity matrices from your multilayer networks.
But the concept of subnetwork can become challenging to interpret, given that it can now include a mix of structural and functional connections.
Andrew
Originally posted by Giuseppe Pontillo:
Dear NBS experts,
I have a cohort of patients for which I obtained both functional and structural networks. My idea is to model the brain connectome as a multilayer network (as in https://pubmed.ncbi.nlm.nih.gov/28327916/) of two layers (SC and FC). Next, the objective would be to test for possible multilayer network differences between patients and a group of HC, as well as to find multilayer subnetworks which correlate with specific behavioral outcomes.
Is there a way to implement the NBS approach with multilayer networks? As far as I understand, it should be possible to use the NBS approach on a 2D representation of the multilayer network (the supra-adjacency matrix). Would it be appropriate?
Thank you,
Giuseppe
I have a cohort of patients for which I obtained both functional and structural networks. My idea is to model the brain connectome as a multilayer network (as in https://pubmed.ncbi.nlm.nih.gov/28327916/) of two layers (SC and FC). Next, the objective would be to test for possible multilayer network differences between patients and a group of HC, as well as to find multilayer subnetworks which correlate with specific behavioral outcomes.
Is there a way to implement the NBS approach with multilayer networks? As far as I understand, it should be possible to use the NBS approach on a 2D representation of the multilayer network (the supra-adjacency matrix). Would it be appropriate?
Thank you,
Giuseppe