Predictive neural signature of internet gaming disorder severity revealed by cross-network connectivity

Background While internet gaming disorder (IGD) correlates with regional brain responses and functional connectivity, however, the brain network architecture underlying addiction severity remains poorly characterized. Methods Using resting-state fMRI data and addiction severity metrics from 586 participants (443 IGD, 143 recreational game users), we employed connectome-based predictive modeling (CPM) with leave-one-out cross-validation to identify neural networks predictive of IGD severity. The resulting network was evaluated for replicability in independent datasets, with key predictive networks and nodes further analyzed. Results CPM identified a replicable addiction severity network. The CPM significantly predicted individual gaming addiction scores (r = 0.19, p < 0.001), with features selected using a threshold of p < 0.01. Predictive power primarily derived from internetwork connectivity linking the subcortical, subvisual, and frontoparietal networks. Validation in independent data showed a directional trend (r=0.17, p=0.011). Conclusions Individual variability in subcortical-subvisual-frontoparietal network connectivity predicts IGD addiction severity, highlighting these circuits as potential targets for neuromodulation interventions.