The field of graph modeling and neuroimaging analysis is rapidly advancing with the development of hierarchical graph transformers and attention mechanisms. These innovations enable the modeling of complex brain networks and the analysis of neuroimaging data with increased accuracy and interpretability. Recent research has focused on designing graph transformers that can capture both local and long-range interactions between brain regions, as well as the hierarchical structure of brain networks. This has led to significant improvements in disease identification and diagnosis, such as depression classification. Additionally, the application of graph transformers to other domains, such as ice layer thickness prediction, has shown promising results. Noteworthy papers include BrainHGT, which proposes a hierarchical graph transformer for interpretable brain network analysis, and NH-GCAT, which introduces a neurocircuitry-inspired hierarchical graph causal attention network for explainable depression identification. These papers demonstrate the potential of hierarchical graph modeling and attention mechanisms to advance our understanding of complex systems and improve diagnosis and treatment of diseases.