Robustness Analysis Of Brain Functional Hypernetwork Based On Sparse Linear Regression Model

The application of brain network analysis in the field of neuroimaging has made some progress.However,the traditional functional connectivity network usually constructs the second-order relationship between brain regions based on pairwise correlation,ignoring the high-order relationship between brain regions.This high-order information may be very important for exploring the intrinsic pathological mechanism of mental illness.In order to represent this high-order information,a super network construction method is proposed.Hypernetwork is defined by hypergraph.Nodes represent brain regions,and hyperedges represent the interaction between multiple brain regions.As a kind of dynamic behavior,robustness is also a research hotspot in the field of hypernetworks,which has important practical significance for building robust networks.It reflects the tolerance of network models to attacks.In human brain network,robustness can simulate the dysfunction of brain disease.Although there are more and more researches on hypernetwork,the dynamic research is relatively less,especially in the field of neuroimaging.In the existing researches on brain functional hypernetworks,most of them focus on the static topological properties of the network,and there is no related research on the dynamic characteristics robustness of brain functional hypernetworks.In order to solve these problems,Lasso,group Lasso and sparse group Lasso methods are introduced to solve the sparse linear regression model to construct the super network,and the structures of the three super networks are compared.Then,based on the two experimental models of node degree and node betweenness attack in intentional attack,the global efficiency and the relative size of the most connected subgraph are used to explore the robustness of the network with node failure in response to attack.The robustness of the network is compared with that of the traditional brain function network to explore a more stable network,and the robustness difference between depression and normal hypernetwork is also studied.The experimental results show that there are differences among the three methods.When constructing super edge,Lasso method is the most strict,group Lasso method is the most relaxed,and sparse group Lasso method is in the middle.In the intentional attack mode,the super network constructed by group Lasso and sparse group Lasso is more robust.At the same time,the super network constructed by group Lasso method is more robust and stable.In the comparative analysis of the robustness of hypernetwork and ordinary network,whether the patients with depression or the normal control group,the brain topology changes of hypernetwork based on sparse linear model with damaged brain nodes are more stable than the traditional brain network,and it has stronger robustness under target attack.In addition,compared with the normal control group,the brain functional super network of depression group was more uniform and randomized,and the network connectivity and network integrity were higher in response to deliberate attack.The hypernetwork model can reflect the direct interaction between human brain more truly and construct a more real and effective brain network,which is very important for more accurate research on the robustness of brain network.The tolerance of the super network to attack,that is,the robustness of the network is very important to evaluate whether the hyper network is stable and reliable.Based on this,this topic has certain research significance and value.