Alterations in the resting-state functional networks are associated with food addiction: an EEG study

Food addiction is considered a clinical and neurobiological overlap between excessive food intake and addictive disorders, yet its underlying brain network mechanisms remain poorly understood. This study aimed to identify the functional network alterations associated with food addiction.

Fifty participants, including individuals with food addiction (FAs, n = 21, mean age = 27.10) and healthy controls (HCs, n = 29, mean age = 31.76), were recruited. Food addiction was identified by the Yale Food Addiction Scale Version 2.0 (YFAS 2.0). Resting-state electroencephalography recordings of both FAs and HCs were evaluated for three types of network features, namely, network connectivity, network properties, and spatial pattern of the network (SPN), to characterize the brain network mechanisms comprehensively. A support vector machine classifier was then employed to evaluate the ability of these network features to distinguish FAs from HCs.

Compared with HCs, FAs demonstrated significantly reduced frontal-parietal connectivity in the alpha band. They also presented altered network properties, including decreased clustering coefficient, global and local efficiency, increased characteristic path length, and distinct SPN features within the alpha band. Partial-correlation network analyses further revealed that alpha-band frontal-parietal connectivity bridged these resting-state network features with YFAS scores. Moreover, a support vector machine classifier integrating alpha-band frontal-parietal connectivity and SPN features achieved a classification accuracy of 92% in distinguishing FAs from HCs.

These findings suggest that alterations in alpha-band frontal-parietal networks may underlie neural differences associated with food addiction, supporting the potential of resting-state brain network patterns to facilitate its diagnosis. Our study provides the first piece of evidence showing that resting-state EEG functional connectivity can distinguish participants with food addiction (FAs) from healthy controls (HCs), and the alterations in alpha-band frontal-parietal networks may be a primary source of impaired neural activity in food addiction. Specifically, we first found that FAs exhibited significantly lower frontal-parietal connectivity, distinctive network properties, and spatial pattern of the network (SPN) parameters. We then utilized a regularized estimation method known as the extended Bayesian information criterion graphical least absolute shrinkage and selection operator (EBICglasso) to explore the complex relationships between multiple clinical scales and brain network features. It revealed that the frontal-parietal connectivity strength bridged the connection between the resting-state EEG network features and YFAS scores. Most importantly, the classification accuracy of feature fusion based on the frontal-parietal connectivity and the SPN reached 96% in distinguishing between FAs and HCs.