Internet addiction (IA) refers to excessive internet use that causes cognitive impairment or distress. Understanding the neurophysiological mechanisms underpinning IA is crucial for enabling an accurate diagnosis and informing treatment and prevention strategies. Despite the recent increase in studies examining the neurophysiological traits of IA, their findings often vary. To enhance the accuracy of identifying key neurophysiological characteristics of IA, this study used the phase lag index (PLI) and weighted PLI (WPLI) methods, which minimize volume conduction effects, to analyze the resting-state electroencephalography (EEG) functional connectivity. We further evaluated the reliability of the identified features for IA classification using various machine learning methods.

Ninety-two participants (42 with IA and 50 healthy controls (HCs)) were included. PLI and WPLI values for each participant were computed, and values exhibiting significant differences between the two groups were selected as features for the subsequent classification task.

Support vector machine (SVM) achieved an 83% accuracy rate using PLI features and an improved 86% accuracy rate using WPLI features. t-test results showed analogous topographical patterns for both the WPLI and PLI. Numerous connections were identified within the delta and gamma frequency bands that exhibited significant differences between the two groups, with the IA group manifesting an elevated level of phase synchronization.

Functional connectivity analysis and machine learning algorithms can jointly distinguish participants with IA from HCs based on EEG data. PLI and WPLI have substantial potential as biomarkers for identifying the neurophysiological traits of IA.

Although aberrant brain regional responses are reported in social anxiety disorder (SAD), little is known about resting-state functional connectivity at the macroscale network level. This study aims to identify functional network abnormalities using a multivariate data-driven method in a relatively large and homogenous sample of SAD patients, and assess their potential diagnostic value.

Forty-six SAD patients and 52 demographically-matched healthy controls (HC) were recruited to undergo clinical evaluation and resting-state functional MRI scanning. We used group independent component analysis to characterize the functional architecture of brain resting-state networks (RSNs) and investigate between-group differences in intra-/inter-network functional network connectivity (FNC). Furtherly, we explored the associations of FNC abnormalities with clinical characteristics, and assessed their ability to discriminate SAD from HC using support vector machine analyses.

SAD patients showed widespread intra-network FNC abnormalities in the default mode network, the subcortical network and the perceptual system (i.e. sensorimotor, auditory and visual networks), and large-scale inter-network FNC abnormalities among those high-order and primary RSNs. Some aberrant FNC signatures were correlated to disease severity and duration, suggesting pathophysiological relevance. Furthermore, intrinsic FNC anomalies allowed individual classification of SAD v. HC with significant accuracy, indicating potential diagnostic efficacy.

SAD patients show distinct patterns of functional synchronization abnormalities both within and across large-scale RSNs, reflecting or causing a network imbalance of bottom-up response and top-down regulation in cognitive, emotional and sensory domains. Therefore, this could offer insights into the neurofunctional substrates of SAD.

Although many previous studies reported structural plasticity of the hippocampus and amygdala induced by electroconvulsive therapy (ECT) in major depressive disorder (MDD), yet the exact roles of both areas for antidepressant effects are still controversial.

In the current study, segmentation of amygdala and hippocampal sub-regions was used to investigate the longitudinal changes of volume, the relationship between volume and antidepressant effects, and prediction performances for ECT in MDD patients before and after ECT using two independent datasets.

As a result, MDD patients showed selectively and consistently increased volume in the left lateral nucleus, right accessory basal nucleus, bilateral basal nucleus, bilateral corticoamygdaloid transition (CAT), bilateral paralaminar nucleus of the amygdala, and bilateral hippocampus-amygdala transition area (HATA) after ECT in both datasets, whereas marginally significant increase of volume in bilateral granule cell molecular layer of the head of dentate gyrus, the bilateral head of cornu ammonis (CA) 4, and left head of CA 3. Correlation analyses revealed that increased volume of left HATA was significantly associated with antidepressant effects after ECT. Moreover, volumes of HATA in the MDD patients before ECT could be served as potential biomarkers to predict ECT remission with the highest accuracy of 86.95% and 82.92% in two datasets (The predictive models were trained on Dataset 2 and the sensitivity, specificity and accuracy of Dataset 2 were obtained from leave-one-out-cross-validation. Thus, they were not independent and very likely to be inflated).

These results not only suggested that ECT could selectively induce structural plasticity of the amygdala and hippocampal sub-regions associated with antidepressant effects of ECT in MDD patients, but also provided potential biomarkers (especially HATA) for effectively and timely interventions for ECT in clinical applications.

Support vector machines (SVMs) based on brain-wise functional connectivity (FC) have been widely adopted for single-subject prediction of patients with schizophrenia, but most of them had small sample size. This study aimed to evaluate the performance of SVMs based on a large single-site dataset and investigate the effects of demographic homogeneity and training sample size on classification accuracy.

The resting functional Magnetic Resonance Imaging (fMRI) dataset comprised 220 patients with schizophrenia and 220 healthy controls. Brain-wise FCs was calculated for each participant and linear SVMs were developed for automatic classification of patients and controls. First, we evaluated the SVMs based on all participants and homogeneous subsamples of men, women, younger (18–30 years), and older (31–50 years) participants by 10-fold nested cross-validation. Then, we hold out a fixed test set of 40 participants (20 patients and 20 controls) and evaluated the SVMs based on incremental training sample sizes (N = 40, 80, …, 400).

We found that the SVMs based on all participants had accuracy of 85.05%. The SVMs based on male, female, young, and older participants yielded accuracy of 84.66, 81.56, 80.50, and 86.13%, respectively. Although the SVMs based on older subsamples had better performance than those based on all participants, they generalized poorly to younger participants (77.24%). For incremental training sizes, the classification accuracy increased stepwise from 72.6 to 83.3%, with >80% accuracy achieved with sample size >240.

The findings indicate that SVMs based on a large dataset yield high classification accuracy and establish models using a large sample size with heterogeneous properties are recommended for single subject prediction of schizophrenia.

Tourette disorder (TD), hallmarks of which are motor and vocal tics, has been related to functional abnormalities in large-scale brain networks. Using a fully data driven approach in a prospective, case–control study, we tested the hypothesis that functional connectivity of these networks carries a neural signature of TD. Our aim was to investigate (i) the brain networks that distinguish adult patients with TD from controls, and (ii) the effects of antipsychotic medication on these networks.

Using a multivariate analysis based on support vector machine (SVM), we developed a predictive model of resting state functional connectivity in 48 patients and 51 controls, and identified brain networks that were most affected by disease and pharmacological treatments. We also performed standard univariate analyses to identify differences in specific connections across groups.

SVM was able to identify TD with 67% accuracy (p = 0.004), based on the connectivity in widespread networks involving the striatum, fronto-parietal cortical areas and the cerebellum. Medicated and unmedicated patients were discriminated with 69% accuracy (p = 0.019), based on the connectivity among striatum, insular and cerebellar networks. Univariate approaches revealed differences in functional connectivity within the striatum in patients v. controls, and between the caudate and insular cortex in medicated v. unmedicated TD.

SVM was able to identify a neuronal network that distinguishes patients with TD from control, as well as medicated and unmedicated patients with TD, holding a promise to identify imaging-based biomarkers of TD for clinical use and evaluation of the effects of treatment.

A few former studies suggested that there are partial overlaps in abnormal brain structure and cognitive function between hypochondriasis (HS) and schizophrenia (SZ). But their differences in brain activity and cognitive function were unclear.

Twenty-one HS patients, 23 SZ patients, and 24 healthy controls (HC) underwent resting-state functional magnetic resonance imaging (rs-fMRI) with the regional homogeneity analysis (ReHo), subsequently exploring the relationship between ReHo value and cognitive functions. The support vector machines (SVM) were used on effectiveness evaluation of ReHo for differentiating HS from SZ.

Compared with HC, HS showed significantly increased ReHo values in right middle temporal gyrus (MTG), left inferior parietal lobe (IPL), and right fusiform gyrus (FG), while SZ showed increased ReHo in left insula, decreased ReHo values in right paracentral lobule. Additionally, HS showed significantly higher ReHo values in FG, MTG, and left paracentral lobule, but lower in insula than SZ. The higher ReHo values in insula were associated with worse performance in MATRICS consensus cognitive battery (MCCB) in HS group. SVM analysis showed a combination of the ReHo values in insula and FG was able to satisfactorily distinguish the HS and SZ patients.

Our results suggested that the altered default mode network (DMN), of which abnormal spontaneous neural activity occurs in multiple brain regions, might play a key role in the pathogenesis of HS, and the resting-state alterations of insula are closely related to cognitive dysfunction in HS. Furthermore, the combination of the ReHo in FG and insula was a relatively ideal indicator to distinguish HS from SZ.

To elucidate the mechanisms of how snack foods may induce non-homeostatic food intake, we used resting state functional magnetic resonance imaging (fMRI), as resting state networks can individually adapt to experience after short time exposures. In addition, we used graph theoretical analysis together with machine learning techniques (support vector machine) to identifying biomarkers that can categorize between high-caloric (potato chips) vs. low-caloric (zucchini) food stimulation.

Seventeen healthy human subjects with body mass index (BMI) 19 to 27 underwent 2 different fMRI sessions where an initial resting state scan was acquired, followed by visual presentation of different images of potato chips and zucchini. There was then a 5-minute pause to ingest food (day 1=potato chips, day 3=zucchini), followed by a second resting state scan. fMRI data were further analyzed using graph theory analysis and support vector machine techniques.

Potato chips vs. zucchini stimulation led to significant connectivity changes. The support vector machine was able to accurately categorize the 2 types of food stimuli with 100% accuracy. Visual, auditory, and somatosensory structures, as well as thalamus, insula, and basal ganglia were found to be important for food classification. After potato chips consumption, the BMI was associated with the path length and degree in nucleus accumbens, middle temporal gyrus, and thalamus.

The results suggest that high vs. low caloric food stimulation in healthy individuals can induce significant changes in resting state networks. These changes can be detected using graph theory measures in conjunction with support vector machine. Additionally, we found that the BMI affects the response of the nucleus accumbens when high caloric food is consumed.