Sparse Network And Component Analysis: A Tutorial on Estimating Cross-Domain Networks

AI Summary

Network approaches model psychopathology as interacting symptom networks, increasingly integrating multiple domains such as psychological, physiological, and brain data.
Cross-domain relations are typically weaker and often excluded by standard estimation methods, hindering detection of inter-domain links.
Sparse Network And Component (SNAC) analysis improves cross-domain relation estimation by modelling multi-domain structure; tutorial available in the R Package SNACR.

Multivariate Behav Res. 2026 Nov 15:1-19. doi: 10.1080/00273171.2026.2738738. Online ahead of print.

ABSTRACT

In network approaches to psychopathology, mental disorders arise out of the interactions between symptoms in a network structure. While these approaches were initially focused on individual psychopathological domains (e.g., how are depression symptoms connected), researchers increasingly include multiple domains in empirical analyses (augmenting psychological symptoms with physiological and brain data). Often, in such analyses all variables are analyzed simultaneously, without taking into account that the data originate from different domains. Such multi-domain networks contain both domain-specific and cross-domain relations. Typically, the cross-domain relations are of particular interest because these represent how domains are connected. However, estimating cross-domain relations presents problems since they are typically weaker than their domain-specific counterparts, such that standard estimation algorithms tend to exclude cross-domain relations. Previously, we have proposed that estimation of cross-domain relations can be improved by taking the multi-domain structure of the data into account using Sparse Network And Component (SNAC) analysis. In this paper, we provide a step-by-step tutorial on how to perform such analysis using the novel R Package SNACR.

PMID:43034898 | DOI:10.1080/00273171.2026.2738738

Document this CPD