High-Order interactions

Function to compute and characterize high-order interactions among n brain areas.

Platform	Version tested
Matlab	R2017b

Scripts

• [sinfo,oinfo,red,syn]=high_order(data,n) Main function. Estimate the O-Information, S-Information, Redundancy, and Synergy among n-plets from 'data' with dimensionality (T, N), where N is the number of brain regions or modules, and T is the number of samples. @author: Marilyn Gatica, [email protected]

n must be greater or equal than three. We test from 3-plets up 20-plets, and we recommend truncating all the data to the same number of samples T.

• [gaussian_data,covmat]=data2gaussian(data) Transform' data' (T samples x N dimension matrix) to Gaussian with mean 0 and standard deviation 1 using empirical copulas. Return the covariance matrix 'covmat' and the transformed gaussian variables 'gaussian_data' @author: Rubén Herzog, [email protected]

• [oinfo,sinfo] = soinfo_from_covmat(covmat,T) Estimate O-Information and S-Information from the covariance matrix 'covmat'. The estimations include analytic bias correction (gaussian_ent_biascorr(N,T)) for the entropy of Gaussian variables depending on 'T' (samples) and 'N' (dimension of 'covmat' matrix). @author: Rubén Herzog and modified by Marilyn Gatica.

Example

We attach an example example1 (without data) to show how to use higher order() among three tiem series. In particular, the example considers 8 subjects and 20 modules.

Questions / comments : [email protected]