The study by Robinaugh, Millner, and McNally explores the identification of highly influential nodes in the network of complicated grief (CG) symptoms. They propose two new indices, one-step expected influence (EI1) and two-step expected influence (EI2), to account for the presence of negative edges, which are often overlooked in traditional centrality measures. Centrality indices, such as strength, closeness, and betweenness, are commonly used to identify influential nodes, but they do not distinguish between positive and negative edges, potentially leading to inaccurate assessments of a node's influence. In simulated networks with exclusively positive edges, centrality and expected influence indices were strongly correlated with observed node influence. However, in networks with negative edges, expected influence indices outperformed centrality indices in predicting observed node influence. The study then applied these indices to an empirical CG network derived from a longitudinal study of bereavement, finding that changes in nodes with high expected influence were more strongly associated with changes in the overall CG network compared to nodes with low expected influence. The findings suggest that high expected influence nodes, such as emotional pain and feelings of emptiness, may play a crucial role in the development and maintenance of CG. This information could inform the design of more effective treatments by targeting these nodes for intervention. However, the study also highlights the limitations of centrality measures in networks with negative edges and the need for further research to refine these indices and understand their implications for CG treatment.The study by Robinaugh, Millner, and McNally explores the identification of highly influential nodes in the network of complicated grief (CG) symptoms. They propose two new indices, one-step expected influence (EI1) and two-step expected influence (EI2), to account for the presence of negative edges, which are often overlooked in traditional centrality measures. Centrality indices, such as strength, closeness, and betweenness, are commonly used to identify influential nodes, but they do not distinguish between positive and negative edges, potentially leading to inaccurate assessments of a node's influence. In simulated networks with exclusively positive edges, centrality and expected influence indices were strongly correlated with observed node influence. However, in networks with negative edges, expected influence indices outperformed centrality indices in predicting observed node influence. The study then applied these indices to an empirical CG network derived from a longitudinal study of bereavement, finding that changes in nodes with high expected influence were more strongly associated with changes in the overall CG network compared to nodes with low expected influence. The findings suggest that high expected influence nodes, such as emotional pain and feelings of emptiness, may play a crucial role in the development and maintenance of CG. This information could inform the design of more effective treatments by targeting these nodes for intervention. However, the study also highlights the limitations of centrality measures in networks with negative edges and the need for further research to refine these indices and understand their implications for CG treatment.