Traditional psychiatry assumes symptoms are caused by an underlying latent disease entity. In contrast, the network approach posits that mental disorders are emergent properties of mutually interacting symptoms. For continuous data, we typically estimate a Gaussian Graphical Model (GGM) where edges represent partial correlations.

To avoid false-positive edges, networks are often estimated using LASSO regularization (Least Absolute Shrinkage and Selection Operator). The glasso algorithm minimizes the objective function: $$\arg\min_{\Theta} \left( \text{tr}(S\Theta) - \log \det(\Theta) + \lambda \sum_{i \neq j} |\theta_{ij}| \right)$$ where $\Theta$ is the precision matrix, $S$ is the sample covariance matrix, and $\lambda$ is the tuning parameter controlling sparsity.