Effective hyper-connectivity network construction and learning: Application to major depressive disorder identification

Jingyu Liu, Wenxin Yang, Yulan Ma, Qunxi Dong, Yang Li, Bin Hu

Abstract: Highlights•We propose a novel effective hyper-connectivity (EHC) network modeling approach that effectively captures the directional information flow among multiple ROIs.•Our proposed directed hypergraph convolutional network (DHGCN) to learn deep representations of brain activity from EHC network and multiple functional indicators of ROIs.•Our framework successfully identifies discriminative EHC that hold potential as clinical biomarker for MDD identification.