Simplifying Multimodal With Single EOG Modality for Automatic Sleep Staging.

Polysomnography (PSG) recordings have been widely used for sleep staging in clinics, containing multiple modality signals (i.e., EEG and EOG). Recently, many studies have combined EEG and EOG modalities for sleep staging, since they are the most and the second most powerful modality for sleep staging among PSG recordings, respectively. However, EEG is complex to collect and sensitive to environment noise or other body activities, imbedding its use in clinical practice. Comparatively, EOG is much more easily to be obtained. In order to make full use of the powerful ability of EEG and the easy collection of EOG, we propose a novel framework to simplify multimodal sleep staging with a single EOG modality. It still performs well with only EOG modality in the absence of the EEG. Specifically, we first model the correlation between EEG and EOG, and then based on the correlation we generate multimodal features with time and frequency guided generators by adopting the idea of generative adversarial learning. We collected a real-world sleep dataset containing 67 recordings and used other four public datasets for evaluation. Compared with other existing sleep staging methods, our framework performs the best when solely using the EOG modality. Moreover, under our framework, EOG provides a comparable performance to EEG.

Gestational diabetes and risk of perinatal depression in low- and middle-income countries: a meta-analysis.

Background: The relationship between gestational diabetes (GDM) and the risk of depression has been thoroughly investigated in high-income countries on their financial basis, while it is largely unexplored in low- and middle- income countries. This meta-analysis aims to assess how GDM influences the risk of perinatal depression by searching multiple electronic databases for studies measuring the odds ratios between them in low- and middle-income countries. Methods: Two independent reviewers searched multiple electronic databases for studies that investigated GDM and perinatal mental disorders on August 31, 2023. Pooled odds ratios (ORs) and confidence intervals (CIs) were calculated using the random effect model. Subgroup analyses were further conducted based on the type of study design and country income level. Results: In total, 16 observational studies met the inclusion criteria. Only the number of studies on depression (n=10) satisfied the conditions to conduct a meta-analysis, showing the relationship between mental illness and GDM has been overlooked in low- and middle-income countries. Evidence shows an elevated risk of perinatal depression in women with GDM (pooled OR 1.92; 95% CI 1.24, 2.97; 10 studies). The increased risk of perinatal depression in patients with GDM was not significantly different between cross-sectional and prospective design. Country income level is a significant factor that adversely influences the risk of perinatal depression in GDM patients. Conclusion: Our findings suggested that women with GDM are vulnerable to perinatal depressive symptoms, and a deeper understanding of potential risk factors and mechanisms may help inform strategies aimed at prevention of exposure to these complications during pregnancy.

DiffMDD: A Diffusion-Based Deep Learning Framework for MDD Diagnosis Using EEG.

Major Depression Disorder (MDD) is a common yet destructive mental disorder that affects millions of people worldwide. Making early and accurate diagnosis of it is very meaningful. Recently, EEG, a non-invasive technique of recording spontaneous electrical activity of brains, has been widely used for MDD diagnosis. However, there are still some challenges in data quality and data size of EEG: (1) A large amount of noise is inevitable during EEG collection, making it difficult to extract discriminative features from raw EEG; (2) It is difficult to recruit a large number of subjects to collect sufficient and diverse data for model training. Both of the challenges cause the overfitting problem, especially for deep learning methods. In this paper, we propose DiffMDD, a diffusion-based deep learning framework for MDD diagnosis using EEG. Specifically, we extract more noise-irrelevant features to improve the model's robustness by designing the Forward Diffusion Noisy Training Module. Then we increase the size and diversity of data to help the model learn more generalized features by designing the Reverse Diffusion Data Augmentation Module. Finally, we re-train the classifier on the augmented dataset for MDD diagnosis. We conducted comprehensive experiments to test the overall performance and each module's effectiveness. The framework was validated on two public MDD diagnosis datasets, achieving the state-of-the-art performance.

Narcolepsy Diagnosis With Sleep Stage Features Using PSG Recordings.

Narcolepsy is a sleep disorder affecting millions of people worldwide and causes serious public health problems. It is hard for doctors to correctly and objectively diagnose narcolepsy. Polysomnography (PSG) recordings, a gold standard for sleep monitoring and quality measurement, can provide abundant and objective cues for the narcolepsy diagnosis. There have been some studies on automatic narcolepsy diagnosis using PSG recordings. However, the sleep stage information, an important cue for narcolepsy diagnosis, has not been fully utilized. For example, some studies have not considered the sleep stage information to diagnose narcolepsy. Although some studies consider the sleep stage information, the stages are manually scored by experts, which is time-consuming and subjective. And the framework using sleep stages scored automatically for narcolepsy diagnosis is designed in a two-phase learning manner, where sleep staging in the first phase and diagnosis in the second phase, causing cumulative error and degrading the performance. To address these challenges, we propose a novel end-to-end framework for automatic narcolepsy diagnosis using PSG recordings. In particular, adopting the idea of multi-task learning, we take the sleep staging as our auxiliary task, and then combine the sleep stage related features with narcolepsy related features for our primary task of narcolepsy diagnosis. We collected a dataset of PSG recordings from 77 participants and evaluated our framework on it. Both of the sleep stage features and the end-to-end fashion contribute to diagnosis performance. Moreover, we do a comprehensive analysis on the relationship between sleep stages and narcolepsy, correlation of different channels, predictive ability of different sensing data, and diagnosis results in subject level.

Spatio-temporal evolution of human neural activity during visually cued hand movements.

Making hand movements in response to visual cues is common in daily life. It has been well known that this process activates multiple areas in the brain, but how these neural activations progress across space and time remains largely unknown. Taking advantage of intracranial electroencephalographic (iEEG) recordings using depth and subdural electrodes from 36 human subjects using the same task, we applied single-trial and cross-trial analyses to high-frequency iEEG activity. The results show that the neural activation was widely distributed across the human brain both within and on the surface of the brain, and focused specifically on certain areas in the parietal, frontal, and occipital lobes, where parietal lobes present significant left lateralization on the activation. We also demonstrate temporal differences across these brain regions. Finally, we evaluated the degree to which the timing of activity within these regions was related to sensory or motor function. The findings of this study promote the understanding of task-related neural processing of the human brain, and may provide important insights for translational applications.

Assisting schizophrenia diagnosis using clinical electroencephalography and interpretable graph neural networks: a real-world and cross-site study.

Schizophrenia (SCZ) is a chronic and serious mental disorder with a high mortality rate. At present, there is a lack of objective, cost-effective and widely disseminated diagnosis tools to address this mental health crisis globally. Clinical electroencephalogram (EEG) is a noninvasive technique to measure brain activity with high temporal resolution, and accumulating evidence demonstrates that clinical EEG is capable of capturing abnormal SCZ neuropathology. Although EEG-based automated diagnostic tools have obtained impressive performance on individual datasets, the transportability of potential EEG biomarkers in cross-site real-world application is still an open question. To address the challenges of small sample sizes and population heterogeneity, we develop an advanced interpretable deep learning model using multimodal clinical EEG features and demographic information as inputs to graph neural networks, and further propose different transfer learning strategies to adapt to different clinical scenarios. Taking the disease discrimination of health control (HC) and SCZ with 1030 participants as a use case, our model is trained on a small clinical dataset (N = 188, Chinese) and enhanced using a large-scale public dataset (N = 508, American) of adult participants. Cross-site validation from an independent dataset of adult participants (N = 157, Chinese) produced stable performance, with AUCs of 0.793-0.852 and accuracies of 0.786-0.858 for different SCZ prevalence, respectively. In addition, cross-site validation from another dataset of adolescent boys (N = 84, Russian) yielded an AUC of 0.702 and an accuracy of 0.690. Moreover, feature visualization further revealed that the ranking of feature importance varied significantly among different datasets, and that EEG theta and alpha band power appeared to be the most significant and translational biomarkers of SCZ pathology. Overall, our promising results demonstrate the feasibility of SCZ discrimination using EEG biomarkers in multiple clinical settings.

An integrated resource for functional and structural connectivity of the marmoset brain.

Comprehensive integration of structural and functional connectivity data is required to model brain functions accurately. While resources for studying the structural connectivity of non-human primate brains already exist, their integration with functional connectivity data has remained unavailable. Here we present a comprehensive resource that integrates the most extensive awake marmoset resting-state fMRI data available to date (39 marmoset monkeys, 710 runs, 12117 mins) with previously published cellular-level neuronal tracing data (52 marmoset monkeys, 143 injections) and multi-resolution diffusion MRI datasets. The combination of these data allowed us to (1) map the fine-detailed functional brain networks and cortical parcellations, (2) develop a deep-learning-based parcellation generator that preserves the topographical organization of functional connectivity and reflects individual variabilities, and (3) investigate the structural basis underlying functional connectivity by computational modeling. This resource will enable modeling structure-function relationships and facilitate future comparative and translational studies of primate brains.

Interictal SEEG Resting-State Connectivity Localizes the Seizure Onset Zone and Predicts Seizure Outcome.

Localization of epileptogenic zone currently requires prolonged intracranial recordings to capture seizure, which may take days to weeks. The authors developed a novel method to identify the seizure onset zone (SOZ) and predict seizure outcome using short-time resting-state stereotacticelectroencephalography (SEEG) data. In a cohort of 27 drug-resistant epilepsy patients, the authors estimated the information flow via directional connectivity and inferred the excitation-inhibition ratio from the 1/f power slope. They hypothesized that the antagonism of information flow at multiple frequencies between SOZ and non-SOZ underlying the relatively stable epilepsy resting state could be related to the disrupted excitation-inhibition balance. They found flatter 1/f power slope in non-SOZ regions compared to the SOZ, with dominant information flow from non-SOZ to SOZ regions. Greater differences in resting-state information flow between SOZ and non-SOZ regions are associated with favorable seizure outcome. By integrating a balanced random forest model with resting-state connectivity, their method localized the SOZ with an accuracy of 88% and predicted the seizure outcome with an accuracy of 92% using clinically determined SOZ. Overall, this study suggests that brief resting-state SEEG data can significantly facilitate the identification of SOZ and may eventually predict seizure outcomes without requiring long-term ictal recordings.

Disrupted fronto-parietal network and default-mode network gamma interactions distinguishing suicidal ideation and suicide attempt in depression.

BACKGROUND: Precise suicide risk evaluation struggled in Major depressive disorder (MDD), especially for patients with only suicidal-ideation (SI) but without suicide attempt (SA). MDD patients have deficits in negative emotion processing, which is associated with the generation of SI and SA. Given the critical role of gamma oscillations in negative emotion processing, we hypothesize that the transition from SI to SA in MDD could be characterized by abnormal gamma interactions. METHODS: We recruited 162 participants containing 106 MDD patients and 56 healthy controls (HCs). Participants performed facial recognition tasks while magnetoencephalography data were recorded. Time-frequency-representation (TFR) analysis was conducted to identify the dominant spectra differences between MDD and HCs, and then source analysis was applied to localize the region of interests. Furthermore, frequency-specific functional connectivity network were constructed and a semi-supervised clustering algorithm was utilized to predict potential suicide risk. RESULTS: Gamma (50-70 Hz) power was found significantly increased in MDD, mainly residing in regions from fronto-parietal-control-network (FPN), visual-network (VN), default-mode-network (DMN) and salience-network (SN). Based on impaired gamma functional connectivity network between well-established SA group and non-SI group, semi-supervised algorithm clustered patients with only SI into two groups with different suicide risks. Moreover, Inter-network gamma connectivity between FPN and DMN significantly negatively correlated with suicide risk and not confounded by depression severity. CONCLUSION: Inter-network gamma connectivity with FPN and DMN might be the key neuropathological interactions underling the progression from SI to SA. By applying semi-supervised clustering to electrophysiological data, it is possible to predict individual suicide risk.

High-Frequency Hubs of the Ictal Cross-Frequency Coupling Network Predict Surgical Outcome in Epilepsy Patients.

Seizure generation is thought to be a process driven by epileptogenic networks; thus, network analysis tools can help determine the efficacy of epilepsy treatment. Studies have suggested that low-frequency (LF) to high-frequency (HF) cross-frequency coupling (CFC) is a useful biomarker for localizing epileptogenic tissues. However, it remains unclear whether the LF or HF coordinated CFC network hubs are more critical in determining the treatment outcome. We hypothesize that HF hubs are primarily responsible for seizure dynamics. Stereo-electroencephalography (SEEG) recordings of 36 seizures from 16 intractable epilepsy patients were analyzed. We propose a new approach to model the temporal-spatial-spectral dynamics of CFC networks. Graph measures are then used to characterize the HF and LF hubs. In the patient group with Engel Class (EC) I outcome, the strength of HF hubs was significantly higher inside the resected regions during the early and middle stages of seizure, while such a significant difference was not observed in the EC III group and only for the early stage in the EC II group. For the LF hubs, a significant difference was identified at the late stage and only in the EC I group. Our findings suggest that HF hubs increase at early and middle stages of the ictal interval while LF hubs increase activity at the late stages. In addition, HF hubs can predict treatment outcomes more precisely, compared to the LF hubs of the CFC network. The proposed method promises to identify more accurate targets for surgical interventions or neuromodulation therapies.

Noninvasive high-frequency oscillations riding spikes delineates epileptogenic sources.

High-frequency oscillations (HFOs) are a promising biomarker for localizing epileptogenic brain and guiding successful neurosurgery. However, the utility and translation of noninvasive HFOs, although highly desirable, is impeded by the difficulty in differentiating pathological HFOs from nonepileptiform high-frequency activities and localizing the epileptic tissue using noninvasive scalp recordings, which are typically contaminated with high noise levels. Here, we show that the consistent concurrence of HFOs with epileptiform spikes (pHFOs) provides a tractable means to identify pathological HFOs automatically, and this in turn demarks an epileptiform spike subgroup with higher epileptic relevance than the other spikes in a cohort of 25 temporal epilepsy patients (including a total of 2,967 interictal spikes and 1,477 HFO events). We found significant morphological distinctions of HFOs and spikes in the presence/absence of this concurrent status. We also demonstrated that the proposed pHFO source imaging enhanced localization of epileptogenic tissue by 162% (∼5.36 mm) for concordance with surgical resection and by 186% (∼12.48 mm) with seizure-onset zone determined by invasive studies, compared to conventional spike imaging, and demonstrated superior congruence with the surgical outcomes. Strikingly, the performance of spike imaging was selectively boosted by the presence of spikes with pHFOs, especially in patients with multitype spikes. Our findings suggest that concurrent HFOs and spikes reciprocally discriminate pathological activities, providing a translational tool for noninvasive presurgical diagnosis and postsurgical evaluation in vulnerable patients.

Frontolimbic alpha activity tracks intentional rest BCI control improvement through mindfulness meditation.

Brain-computer interfaces (BCIs) are capable of translating human intentions into signals controlling an external device to assist patients with severe neuromuscular disorders. Prior work has demonstrated that participants with mindfulness meditation experience evince improved BCI performance, but the underlying neural mechanisms remain unclear. Here, we conducted a large-scale longitudinal intervention study by training participants in mindfulness-based stress reduction (MBSR; a standardized mind-body awareness training intervention), and investigated whether and how short-term MBSR affected sensorimotor rhythm (SMR)-based BCI performance. We hypothesize that MBSR training improves BCI performance by reducing mind wandering and enhancing self-awareness during the intentional rest BCI control, which would mainly be reflected by modulations of default-mode network and limbic network activity. We found that MBSR training significantly improved BCI performance compared to controls and these behavioral enhancements were accompanied by increased frontolimbic alpha activity (9-15 Hz) and decreased alpha connectivity among limbic network, frontoparietal network, and default-mode network. Furthermore, the modulations of frontolimbic alpha activity were positively correlated with the duration of meditation experience and the extent of BCI performance improvement. Overall, these data suggest that mindfulness allows participant to reach a state where they can modulate frontolimbic alpha power and improve BCI performance for SMR-based BCI control.

Mindfulness Improves Brain-Computer Interface Performance by Increasing Control Over Neural Activity in the Alpha Band.

Brain-computer interfaces (BCIs) are promising tools for assisting patients with paralysis, but suffer from long training times and variable user proficiency. Mind-body awareness training (MBAT) can improve BCI learning, but how it does so remains unknown. Here, we show that MBAT allows participants to learn to volitionally increase alpha band neural activity during BCI tasks that incorporate intentional rest. We trained individuals in mindfulness-based stress reduction (MBSR; a standardized MBAT intervention) and compared performance and brain activity before and after training between randomly assigned trained and untrained control groups. The MBAT group showed reliably faster learning of BCI than the control group throughout training. Alpha-band activity in electroencephalogram signals, recorded in the volitional resting state during task performance, showed a parallel increase over sessions, and predicted final BCI performance. The level of alpha-band activity during the intentional resting state correlated reliably with individuals' mindfulness practice as well as performance on a breath counting task. Collectively, these results show that MBAT modifies a specific neural signal used by BCI. MBAT, by increasing patients' control over their brain activity during rest, may increase the effectiveness of BCI in the large population who could benefit from alternatives to direct motor control.

Discriminating Suicide Attempters and Predicting Suicide Risk Using Altered Frontolimbic Resting-State Functional Connectivity in Patients With Bipolar II Disorder.

Bipolar II disorder (BD-II) major depression episode is highly associated with suicidality, and objective neural biomarkers could be key elements to assist in early prevention and intervention. This study aimed to integrate altered brain functionality in the frontolimbic system and machine learning techniques to classify suicidal BD-II patients and predict suicidality risk at the individual level. A cohort of 169 participants were enrolled, including 43 BD-II depression patients with at least one suicide attempt during a current depressive episode (SA), 62 BD-II depression patients without a history of attempted suicide (NSA), and 64 demographically matched healthy controls (HCs). We compared resting-state functional connectivity (rsFC) in the frontolimbic system among the three groups and explored the correlation between abnormal rsFCs and the level of suicide risk (assessed using the Nurses' Global Assessment of Suicide Risk, NGASR) in SA patients. Then, we applied support vector machines (SVMs) to classify SA vs. NSA in BD-II patients and predicted the risk of suicidality. SA patients showed significantly decreased frontolimbic rsFCs compared to NSA patients. The left amygdala-right middle frontal gyrus (orbital part) rsFC was negatively correlated with NGASR in the SA group, but not the severity of depressive or anxiety symptoms. Using frontolimbic rsFCs as features, the SVMs obtained an overall 84% classification accuracy in distinguishing SA and NSA. A significant correlation was observed between the SVMs-predicted NGASR and clinical assessed NGASR (r = 0.51, p = 0.001). Our results demonstrated that decreased rsFCs in the frontolimbic system might be critical objective features of suicidality in BD-II patients, and could be useful for objective prediction of suicidality risk in individuals.

Structural-functional decoupling predicts suicide attempts in bipolar disorder patients with a current major depressive episode.

Bipolar disorder (BD) is associated with a high risk of suicidality, and it is challenging to predict suicide attempts in clinical practice to date. Although structural and functional connectivity alterations from neuroimaging studies have been previously reported in BD with suicide attempts, little is known about how abnormal structural and functional connectivity relates to each other. Here, we hypothesize that structure connectivity constrains functional connectivity, and structural-functional coupling is a more sensitive biomarker to detect subtle brain abnormalities than any single modality in BD patients with a current major depressive episode who had attempted suicide. By investigating structural and resting-state fMRI connectivity, as well as their coupling among 191 BD depression patients with or without a history of suicide attempts and 113 healthy controls, we found that suicide attempters in BD depression patients showed significantly decreased central-temporal structural connectivity, increased frontal-temporal functional connectivity, along with decreased structural-functional coupling compared with non-suicide attempters. Crucially, the altered structural connectivity network predicted the abnormal functional connectivity network profile, and the structural-functional coupling was significantly correlated with suicide risk but not with depression or anxiety severity. Our findings suggest that the structural connectome is the key determinant of brain dysfunction, and structural-functional coupling could serve as a valuable trait-like biomarker for BD suicidal predication over and above the intramodality network connectivity. Such a measure can have clinical implications for early identification of suicide attempters with BD depression and inform strategies for prevention.

Autonomic Nervous System Is Related to Inhibitory and Control Function Through Functional Inter-Region Connectivities of OFC in Major Depression.

OBJECTIVE: To investigate the mechanism of interactions between autonomic nervous system (ANS) and cognitive function in Major depression (MD) with Magnetoencephalography (MEG) measurements. METHODS: Participants with MD (n = 20), and Health controls (HCs, n = 18) were completed MEG measurements during the performance of a go/no-go task. Heart rate variability (HRV) indices (SDANN, and RMSSD) were derived from the raw MEG data. The correlation analysis of the HRV and functional connectivities in different brain regions was conducted by Pearson's r in two groups. RESULTS: The go/no-go task performances of HCs were better than MD patients; HRV indices were lower in the MD group. Under the no-go task, a brain MEG functional connectivity analysis based on the seed regions of the orbitofrontal cortex (OFC) displayed increased functional inter-region connectivity networks of OFC in MD group. HRV indices were correlated with different functional inter-region connectivity networks of OFC in two groups, respectively. CONCLUSION: ANS is related to inhibitory and control function through functional inter-region connectivity networks of OFC in MD. These findings have important implications for the understanding pathophysiology of MD, and MEG may provide an image-guided tool for interventions.

Brain-Heart Interactions Underlying Traditional Tibetan Buddhist Meditation.

Despite accumulating evidence suggesting improvement in one's well-being as a result of meditation, little is known about if or how the brain and the periphery interact to produce these behavioral and mental changes. We hypothesize that meditation reflects changes in the neural representations of visceral activity, such as cardiac behavior, and investigated the integration of neural and visceral systems and the spontaneous whole brain spatiotemporal dynamics underlying traditional Tibetan Buddhist meditation. In a large cohort of long-term Tibetan Buddhist monk meditation practitioners, we found distinct transient modulations of the neural response to heartbeats in the default mode network (DMN), along with large-scale network reconfigurations in the gamma and theta bands of electroencephalography (EEG) activity induced by meditation. Additionally, temporal-frontal network connectivity in the EEG theta band was negatively correlated with the duration of meditation experience, and gamma oscillations were uniquely, directionally coupled to theta oscillations during meditation. Overall, these data suggest that the neural representation of cardiac activity in the DMN and large-scale spatiotemporal network integrations underlie the fundamental neural mechanism of meditation and further imply that meditation may utilize cortical plasticity, inducing both immediate and long-lasting changes in the intrinsic organization and activity of brain networks.

Distinct directional couplings between slow and fast gamma power to the phase of theta oscillations in the rat hippocampus.

It is well-established that theta (~4-10 Hz) and gamma (~25-100 Hz) oscillations interact in the rat hippocampus. This cross-frequency coupling might facilitate neuronal coordination both within and between brain areas. However, it remains unclear whether the phase of theta oscillations controls the power of slow and fast gamma activity or vice versa. We here applied spectral Granger causality, phase slope index and a newly developed cross-frequency directionality (CFD) measure to investigate directional interactions between local field potentials recorded within and across hippocampal subregions of CA1 and CA3 of freely exploring rats. Given the well-known CA3 to CA1 anatomical connection, we hypothesized that interregional directional interactions were constrained by anatomical connection, and within-frequency and cross-frequency directional interactions were always from CA3 to CA1. As expected, we found that CA3 drove CA1 in the theta band, and theta phase-to-gamma power coupling was prominent both within and between CA3 and CA1 regions. The CFD measure further demonstrated that distinct directional couplings with respect to theta phase was different between slow and fast gamma activity. Importantly, CA3 slow gamma power phase-adjusted CA1 theta oscillations, suggesting that slow gamma activity in CA3 entrains theta oscillations in CA1. In contrast, CA3 theta phase controls CA1 fast gamma activity, indicating that communication at CA1 fast gamma is coordinated by CA3 theta phase. Overall, these findings demonstrate dynamic directional interactions between theta and slow/fast gamma oscillations in the hippocampal network, suggesting that anatomical connections constrain the directional interactions.

Magnetoencephalography resting-state spectral fingerprints distinguish bipolar depression and unipolar depression.

OBJECTIVES: In clinical practice, bipolar depression (BD) and unipolar depression (UD) appear to have similar symptoms, causing BD being frequently misdiagnosed as UD, leading to improper treatment decision and outcome. Therefore, it is in urgent need of distinguishing BD from UD based on clinical objective biomarkers as early as possible. Here, we aimed to integrate brain neuroimaging data and an advanced machine learning technique to predict different types of mood disorder patients at the individual level. METHODS: Eyes closed resting-state magnetoencephalography (MEG) data were collected from 23 BD, 30 UD, and 31 healthy controls (HC). Individual power spectra were estimated by Fourier transform, and statistic spectral differences were assessed via a cluster permutation test. A support vector machine classifier was further applied to predict different mood disorder types based on discriminative oscillatory power. RESULTS: Both BD and UD showed decreased frontal-central gamma/beta ratios comparing to HC, in which gamma power (30-75 Hz) was decreased in BD while beta power (14-30 Hz) was increased in UD vs HC. The support vector machine model obtained significant high classification accuracies distinguishing three groups based on mean gamma and beta power (BD: 79.9%, UD: 81.1%, HC: 76.3%, P < .01). CONCLUSIONS: In combination with resting-state MEG data and machine learning technique, it is possible to make an individual and objective prediction for mode disorder types, which in turn has implications for diagnosis precision and treatment decision of mood disorder patients.