Brain Function Analysis Of Insomnia Disorder Based On Hypergraph Combined With Deep Learning

What is it about?

Insomnia (ID) is a chronic sleep disorder with a high prevalence and a heavy socioeconomic burden. Despite extensive research, the relationship between ID and brain activity is complex and the exact etiology is difficult to determine. In this study, we propose a prediction framework for ID categorization. First, a higher order functional connectivity network was constructed using hypergraphs to capture higher-order interactions between multiple brain ROIs. Then, the high- and low-order functional connectivity networks are fused through a self-attentive mechanism to utilize the complementary information of the features. Finally, a spatial-temporally structured adaptive graph convolutional network was used to classify and predict the ID dataset and the normal control (NC) group collected from Qingdao University Hospital with accuracy and AUC of 83.3% and 83.5%, respectively. Compared with other methods, our method shows excellent performance and provides a promising direction for the development of effective diagnostic methods for insomnia.

Featured Image

Photo by Growtika on Unsplash

Photo by Growtika on Unsplash