Resting-state FMRI Features In Predicting Internet Gaming Disorder Severity

As an emerging behavioral addiction,internet gaming disorder(IGD)has become a worldwide mental health concern.Although it has attracted the great interest of many researchers,its neural mechanism is still not clear.Functional magnetic resonance imaging(fMRI)technology is a promising approach to reveal the neural features of IGD,and it can help people further understand the occurrence and development of IGD at the neural level.Previous brain imaging studies typically investigated the neural mechanism of IGD at the group level and the traditional univariate analysis could not effectively utilize massive imaging data,resulting in a waste of data.Machine learning is a data-driven multivariate analysis method and broadly employed in the field of neuroscience.Therefore,the present study aims to use machine learning methods to predict the IGD severity at the individual level by using multiple resting-state fMRI metrics and explore the reliable neural biomarker of IGD.Study 1 included 402 participants with varying IGD severity.Using multivariate prediction analysis,the predictive models based on ReHo and ALFF significantly predicted IGD severity.The right precentral gyrus and the left postcentral gyrus were consensus high-weight brain regions of two predictive models.The results of graph theory analysis found that the nodal properties of the right precentral gyrus were significantly positively correlated with IGD severity,while no significant associations were found between IGD severity and the left postcentral gyrus.Granger causality analysis showed that the effective connectivity(EC)from the right precentral gyrus to the dorsal anterior cingulate cortex(d ACC)was positively correlated with IGD severity.These results suggested that the precentral gyrus is an important region related to IGD.The altered EC from the precentral gyrus to the d ACC may be the underlying neural substrate of IGD.Study 2 applied the same imaging data as Study 1.After constructing the restingstate brain functional connectivity matrix for each participant,the functional connectomes were used to predict IGD severity.The results of connectome-based predictive model showed that the functional connectomes significantly predicted IGD severity.The high-degree nodes in the connectome included the superior temporal sulcus,precentral gyrus,orbitofrontal cortex,middle frontal gyrus and superior frontal gyrus.Besides,the connectome mainly consisted of functional connectivity with the default network,within the motor/sensory network,and between the default network and motor/sensory network.The results suggested that the brain regions related to social cognition and motor coordination play a key role in IGD and emphasized the relationship between the default and motor/sensory network and IGD from the perspective of large-scale brain networks.Study 3 used the same data in Study 1 and 2.Based on graph theory,three node topological metrics were applied as features to predict IGD severity.The results demonstrated that node degree centrality(DC),but not nodal betweenness centrality and nodal efficiency,significantly predicted IGD severity.In the DC-based predictive model,high-weight nodes included precentral gyrus,lingual gyrus,orbitofrontal cortex,posterior cingulate cortex.In addition,we also found that the default mode network and motor/sensory network had relatively higher weights than other brain networks.The results of this study further proved the important role of the precentral gyrus and the orbitofrontal cortex in IGD and supported that view that IGD is mainly related to the default mode network and motor/sensory network.In summary,the resting-state fMRI metrics could significantly predict IGD severity individually.The machine learning methods revealed the brain regions and functional coupling between brain regions that are closely related to IGD.The results indicate that the precentral gyrus may be a potential biomarker of IGD.Meanwhile,the abnormality of the default mode network and motor/sensory network may be the underlying neural substrates of IGD.These findings deepen our understanding in the neural mechanism of IGD and provide valuable reference for the diagnosis,intervention,and treatment of IGD.