Identifying key psychological symptoms by a higher-order network-based approach

The pairwise psychological symptom network approach treats symptoms or questionnaire items as nodes; the pairwise connections between nodes reflect their statistical dependencies and are widely used to identify key symptoms of people with mental disorders. Generally speaking, a person with a mental disorder usually has multiple psychological symptoms, which it is difficult to locate by the pairwise psychological symptom network approach. We propose a higher-order psychological symptom network to describe the coexistence of psychological symptoms for individuals and use nodes to denote different symptoms in psychological scales. Hyperedges represent the relationships among multiple symptoms. We construct the pairwise psychological symptom networks and higher-order psychological symptom networks for the CHARLS, NHANES, and NHRVS databases. The results show that in both pairwise and higher-order symptom networks, node strength shows better stability compared to closeness and betweenness. The key psychological symptoms are consistent across pairwise and higher-order psychological symptom networks, whereas the key co-emergence of multiple symptoms exhibits variability. In addition, there are differences between mentally disordered and mentally healthy populations in the node characterization of higher-order psychological symptom networks. Overall, the higher-order psychological symptom network method proposed in this study extends current psychological symptoms network analysis approaches. It enriches the psychology toolbox and opens new avenues for future research.