Resting-State EEG Signal for Major Depressive Disorder Detection: A Systematic Validation on a Large and Diverse Dataset

Emotion recognition using hierarchical spatial-temporal learning transformer from regional to global brain

Emotion recognition is an essential but challenging task in human-computer interaction systems due to the distinctive spatial structures and dynamic temporal dependencies associated with each emotion. However, current approaches fail to accurately capture the intricate effects of electroencephalogram (EEG) signals across different brain regions on emotion recognition. Therefore, this paper designs a transformer-based method, denoted by R2G-STLT, which relies on a spatial-temporal transformer encoder with regional to global hierarchical learning that learns the representative spatiotemporal features from the electrode level to the brain-region level. The regional spatial-temporal transformer (RST-Trans) encoder is designed to obtain spatial information and context dependence at the electrode level aiming to learn the regional spatiotemporal features. Then, the global spatial-temporal transformer (GST-Trans) encoder is utilized to extract reliable global spatiotemporal features, reflecting the impact of various brain regions on emotion recognition tasks. Moreover, the multi-head attention mechanism is placed into the GST-Trans encoder to empower it to capture the long-range spatial-temporal information among the brain regions. Finally, subject-independent experiments are conducted on each frequency band of the DEAP, SEED, and SEED-IV datasets to assess the performance of the proposed model. Results indicate that the R2G-STLT model surpasses several state-of-the-art approaches.

Resting-State Electroencephalogram Depression Diagnosis Based on Traditional Machine Learning and Deep Learning: A Comparative Analysis

The global prevalence of Major Depressive Disorder (MDD) is increasing at an alarming rate, underscoring the urgent need for timely and accurate diagnoses to facilitate effective interventions and treatments. Electroencephalography remains a widely used neuroimaging technique in psychiatry, due to its non-invasive nature and cost-effectiveness. With the rise of computational psychiatry, the integration of EEG with artificial intelligence has yielded remarkable results in diagnosing depression. This review offers a comparative analysis of two predominant methodologies in research: traditional machine learning and deep learning methods. Furthermore, this review addresses key challenges in current research and suggests potential solutions. These insights aim to enhance diagnostic accuracy for depression and also foster further development in the area of computational psychiatry.

Multimodal Fusion of EEG and Audio Spectrogram for Major Depressive Disorder Recognition Using Modified DenseNet121

BACKGROUND/OBJECTIVES

This study investigates the classification of Major Depressive Disorder (MDD) using electroencephalography (EEG) Short-Time Fourier-Transform (STFT) spectrograms and audio Mel-spectrogram data of 52 subjects. The objective is to develop a multimodal classification model that integrates audio and EEG data to accurately identify depressive tendencies.

METHODS

We utilized the Multimodal open dataset for Mental Disorder Analysis (MODMA) and trained a pre-trained Densenet121 model using transfer learning. Features from both the EEG and audio modalities were extracted and concatenated before being passed through the final classification layer. Additionally, an ablation study was conducted on both datasets separately.

RESULTS

The proposed multimodal classification model demonstrated superior performance compared to existing methods, achieving an Accuracy of 97.53%, Precision of 98.20%, F1 Score of 97.76%, and Recall of 97.32%. A confusion matrix was also used to evaluate the model's effectiveness.

CONCLUSIONS

The paper presents a robust multimodal classification approach that outperforms state-of-the-art methods with potential application in clinical diagnostics for depression assessment.

Resting-state EEG features modulated by depressive state in healthy individuals: insights from theta PSD, theta-beta ratio, frontal-parietal PLV, and sLORETA

Depressive states in both healthy individuals and those with major depressive disorder exhibit differences primarily in symptom severity rather than symptom type, suggesting that there is a spectrum of depressive symptoms. The increasing prevalence of mild depression carries lifelong implications, emphasizing its clinical and social significance, which parallels that of moderate depression. Early intervention and psychotherapy have shown effective outcomes in subthreshold depression. Electroencephalography serves as a non-invasive, powerful tool in depression research, with many studies employing it to discover biomarkers and explore underlying mechanisms for the identification and diagnosis of depression. However, the efficacy of these biomarkers in distinguishing various depressive states in healthy individuals and in understanding the associated mechanisms remains uncertain. In our study, we examined the power spectrum density and the region-based phase-locking value in healthy individuals with various depressive states during their resting state. We found significant differences in neural activity, even among healthy individuals. Participants were categorized into high, middle, and low depressive state groups based on their response to a questionnaire, and eyes-open resting-state electroencephalography was conducted. We observed significant differences among the different depressive state groups in theta- and beta-band power, as well as correlations in the theta-beta ratio in the frontal lobe and phase-locking connections in the frontal, parietal, and temporal lobes. Standardized low-resolution electromagnetic tomography analysis for source localization comparing the differences in resting-state networks among the three depressive state groups showed significant differences in the frontal and temporal lobes. We anticipate that our study will contribute to the development of effective biomarkers for the early detection and prevention of depression.

Power spectral density-based resting-state EEG classification of first-episode psychosis

Historically, the analysis of stimulus-dependent time-frequency patterns has been the cornerstone of most electroencephalography (EEG) studies. The abnormal oscillations in high-frequency waves associated with psychotic disorders during sensory and cognitive tasks have been studied many times. However, any significant dissimilarity in the resting-state low-frequency bands is yet to be established. Spectral analysis of the alpha and delta band waves shows the effectiveness of stimulus-independent EEG in identifying the abnormal activity patterns of pathological brains. A generalized model incorporating multiple frequency bands should be more efficient in associating potential EEG biomarkers with first-episode psychosis (FEP), leading to an accurate diagnosis. We explore multiple machine-learning methods, including random-forest, support vector machine, and Gaussian process classifier (GPC), to demonstrate the practicality of resting-state power spectral density (PSD) to distinguish patients of FEP from healthy controls. A comprehensive discussion of our preprocessing methods for PSD analysis and a detailed comparison of different models are included in this paper. The GPC model outperforms the other models with a specificity of 95.78% to show that PSD can be used as an effective feature extraction technique for analyzing and classifying resting-state EEG signals of psychiatric disorders.

Characterizing Major Depressive Disorder (MDD) using alpha-band activity in resting-state electroencephalogram (EEG) combined with MATRICS Consensus Cognitive Battery (MCCB)

BACKGROUND

The diagnosis of major depressive disorder (MDD) is commonly based on the subjective evaluation by experienced psychiatrists using clinical scales. Hence, it is particularly important to find more objective biomarkers to aid in diagnosis and further treatment. Alpha-band activity (7-13 Hz) is the most prominent component in resting electroencephalogram (EEG), which is also thought to be a potential biomarker. Recent studies have shown the existence of multiple sub-oscillations within the alpha band, with distinct neural underpinnings. However, the specific contribution of these alpha sub-oscillations to the diagnosis and treatment of MDD remains unclear.

METHODS

In this study, we recorded the resting-state EEG from MDD and HC populations in both open and closed-eye state conditions. We also assessed cognitive processing using the MATRICS Consensus Cognitive Battery (MCCB).

RESULTS

We found that the MDD group showed significantly higher power in the high alpha range (10.5-11.5 Hz) and lower power in the low alpha range (7-8.5 Hz) compared to the HC group. Notably, high alpha power in the MDD group is negatively correlated with working memory performance in MCCB, whereas no such correlation was found in the HC group. Furthermore, using five established classification algorithms, we discovered that combining alpha oscillations with MCCB scores as features yielded the highest classification accuracy compared to using EEG or MCCB scores alone.

CONCLUSIONS

Our results demonstrate the potential of sub-oscillations within the alpha frequency band as a potential distinct biomarker. When combined with psychological scales, they may provide guidance relevant for the diagnosis and treatment of MDD.

Selecting EEG channels and features using multi-objective optimization for accurate MCI detection: validation using leave-one-subject-out strategy

Effective management of dementia requires the timely detection of mild cognitive impairment (MCI). This paper introduces a multi-objective optimization approach for selecting EEG channels (and features) for the purpose of detecting MCI. Firstly, each EEG signal from each channel is decomposed into subbands using either variational mode decomposition (VMD) or discrete wavelet transform (DWT). A feature is then extracted from each subband using one of the following measures: standard deviation, interquartile range, band power, Teager energy, Katz's and Higuchi's fractal dimensions, Shannon entropy, sure entropy, or threshold entropy. Different machine learning techniques are used to classify the features of MCI cases from those of healthy controls. The classifier's performance is validated using leave-one-subject-out (LOSO) cross-validation (CV). The non-dominated sorting genetic algorithm (NSGA)-II is designed with the aim of minimizing the number of EEG channels (or features) and maximizing classification accuracy. The performance is evaluated using a publicly available online dataset containing EEGs from 19 channels recorded from 24 participants. The results demonstrate a significant improvement in performance when utilizing the NSGA-II algorithm. By selecting only a few appropriate EEG channels, the LOSO CV-based results show a significant improvement compared to using all 19 channels. Additionally, the outcomes indicate that accuracy can be further improved by selecting suitable features from different channels. For instance, by combining VMD and Teager energy, the SVM accuracy obtained using all channels is 74.24%. Interestingly, when only five channels are selected using NSGA-II, the accuracy increases to 91.56%. The accuracy is further improved to 95.28% when using only 8 features selected from 7 channels. This demonstrates that by choosing informative features or channels while excluding noisy or irrelevant information, the impact of noise is reduced, resulting in improved accuracy. These promising findings indicate that, with a limited number of channels and features, accurate diagnosis of MCI is achievable, which opens the door for its application in clinical practice.

Identifying EEG Biomarkers of Depression with Novel Explainable Deep Learning Architectures

Deep learning methods are increasingly being applied to raw electroencephalogram (EEG) data. However, if these models are to be used in clinical or research contexts, methods to explain them must be developed, and if these models are to be used in research contexts, methods for combining explanations across large numbers of models must be developed to counteract the inherent randomness of existing training approaches. Model visualization-based explainability methods for EEG involve structuring a model architecture such that its extracted features can be characterized and have the potential to offer highly useful insights into the patterns that they uncover. Nevertheless, model visualization-based explainability methods have been underexplored within the context of multichannel EEG, and methods to combine their explanations across folds have not yet been developed. In this study, we present two novel convolutional neural network-based architectures and apply them for automated major depressive disorder diagnosis. Our models obtain slightly lower classification performance than a baseline architecture. However, across 50 training folds, they find that individuals with MDD exhibit higher β power, potentially higher δ power, and higher brain-wide correlation that is most strongly represented within the right hemisphere. This study provides multiple key insights into MDD and represents a significant step forward for the domain of explainable deep learning applied to raw EEG. We hope that it will inspire future efforts that will eventually enable the development of explainable EEG deep learning models that can contribute both to clinical care and novel medical research discoveries.

Diagnostic deep learning algorithms that use resting EEG to distinguish major depressive disorder, bipolar disorder, and schizophrenia from each other and from healthy volunteers

BACKGROUND

Mood disorders and schizophrenia affect millions worldwide. Currently, diagnosis is primarily determined by reported symptomatology. As symptoms may overlap, misdiagnosis is common, potentially leading to ineffective or destabilizing treatment. Diagnostic biomarkers could significantly improve clinical care by reducing dependence on symptomatic presentation.

METHODS

We used deep learning analysis (DLA) of resting electroencephalograph (EEG) to differentiate healthy control (HC) subjects (N = 239), from those with major depressive disorder (MDD) (N = 105), MDD-atypical (MDD-A) (N = 27), MDD-psychotic (MDD-P) (N = 35), bipolar disorder-depressed episode (BD-DE) (N = 71), BD-manic episode (BD-ME) (N = 49), and schizophrenia (SCZ) (N = 122) and also differentiate subjects with mental disorders on a pair-wise basis. DSM-III-R diagnoses were determined and supplemented by computerized Quick Diagnostic Interview Schedule. After EEG preprocessing, robust exact low-resolution electromagnetic tomography (ReLORETA) computed EEG sources for 82 brain regions. 20 % of all subjects were then set aside for independent testing. Feature selection methods were then used for the remaining subjects to identify brain source regions that are discriminating between diagnostic categories.

RESULTS

Pair-wise classification accuracies between 90 % and 100 % were obtained using independent test subjects whose data were not used for training purposes. The most frequently selected features across various pairs are in the postcentral, supramarginal, and fusiform gyri, the hypothalamus, and the left cuneus. Brain sites discriminating SCZ from HC were mainly in the left hemisphere while those separating BD-ME from HC were on the right.

LIMITATIONS

The use of superseded DSM-III-R diagnostic system and relatively small sample size in some disorder categories that may increase the risk of overestimation.

CONCLUSIONS

DLA of EEG could be trained to autonomously classify psychiatric disorders with over 90 % accuracy compared to an expert clinical team using standardized operational methods.

Development of Artificial Intelligence for Determining Major Depressive Disorder Based on Resting-State EEG and Single-Pulse Transcranial Magnetic Stimulation-Evoked EEG Indices

Depression is the disorder with the greatest socioeconomic burdens. Its diagnosis is still based on an operational diagnosis derived from symptoms, and no objective diagnostic indicators exist. Thus, the present study aimed to develop an artificial intelligence (AI) model to aid in the diagnosis of depression from electroencephalography (EEG) data by applying machine learning to resting-state EEG and transcranial magnetic stimulation (TMS)-evoked EEG acquired from patients with depression and healthy controls. Resting-state EEG and single-pulse TMS-EEG were acquired from 60 patients and 60 healthy controls. Power spectrum analysis, phase synchronization analysis, and phase-amplitude coupling analysis were conducted on EEG data to extract feature candidates to apply different types of machine learning algorithms. Furthermore, to address the limitation of the sample size, dimensionality reduction was performed in a manner to increase the quality of information by featuring robust neurophysiological metrics that showed significant differences between the two groups. Then, nine different machine learning models were applied to the data. For the EEG data, we created models combining four modalities, including (1) resting-state EEG, (2) pre-stimulus TMS-EEG, (3) post-stimulus TMS-EEG, and (4) differences between pre- and post-stimulus TMS-EEG, and evaluated their performance. We found that the best estimation performance (a mean area under the curve of 0.922) was obtained using receiver operating characteristic curve analysis when linear discriminant analysis (LDA) was applied to the combination of the four feature sets. This study showed that by using TMS-EEG neurophysiological indices as features, it is possible to develop a depression decision-support AI algorithm that exhibits high discrimination accuracy.

Fractal Geometry Meets Computational Intelligence: Future Perspectives

Characterizations in terms of fractals are typically employed for systems with complex and multiscale descriptions. A prominent example of such systems is provided by the human brain, which can be idealized as a complex dynamical system made of many interacting subunits. The human brain can be modeled in terms of observable variables together with their spatio-temporal-functional relations. Computational intelligence is a research field bridging many nature-inspired computational methods, such as artificial neural networks, fuzzy systems, and evolutionary and swarm intelligence optimization techniques. Typical problems faced by means of computational intelligence methods include those of recognition, such as classification and prediction. Although historically conceived to operate in some vector space, such methods have been recently extended to the so-called nongeometric spaces, considering labeled graphs as the most general example of such patterns. Here, we suggest that fractal analysis and computational intelligence methods can be exploited together in neuroscience research. Fractal characterizations can be used to (i) assess scale-invariant properties and (ii) offer numeric, feature-based representations to complement the usually more complex pattern structures encountered in neurosciences. Computational intelligence methods could be used to exploit such fractal characterizations, considering also the possibility to perform data-driven analysis of nongeometric input spaces, therby overcoming the intrinsic limits related to Euclidean geometry.

Neurosteroids: mechanistic considerations and clinical prospects

Like other classes of treatments described in this issue's section, neuroactive steroids have been studied for decades but have risen as a new class of rapid-acting, durable antidepressants with a distinct mechanism of action from previous antidepressant treatments and from other compounds covered in this issue. Neuroactive steroids are natural derivatives of progesterone but are proving effective as exogenous treatments. The best understood mechanism is that of positive allosteric modulation of GABAA receptors, where subunit selectivity may promote their profile of action. Mechanistically, there is some reason to think that neuroactive steroids may separate themselves from liabilities of other GABA modulators, although research is ongoing. It is also possible that intracellular targets, including inflammatory pathways, may be relevant to beneficial actions. Strengths and opportunities for further development include exploiting non-GABAergic targets, structural analogs, enzymatic production of natural steroids, precursor loading, and novel formulations. The molecular mechanisms of behavioral effects are not fully understood, but study of brain network states involved in emotional processing demonstrate a robust influence on affective states not evident with at least some other GABAergic drugs including benzodiazepines. Ongoing studies with neuroactive steroids will further elucidate the brain and behavioral effects of these compounds as well as likely underpinnings of disease.

Clinical Sensitivity of Fractal Neurodynamics

Among the significant advances in the understanding of the organization of the neuronal networks that coordinate the body and brain, their complex nature is increasingly important, resulting from the interaction between the very large number of constituents strongly organized hierarchically and at the same time with "self-emerging." This awareness drives us to identify the measures that best quantify the "complexity" that accompanies the continuous evolutionary dynamics of the brain. In this chapter, after an introductory section (Sect. 15.1), we examine how the Higuchi fractal dimension is able to perceive physiological processes (15.2), neurological (15.3) and psychiatric (15.4) disorders, and neuromodulation effects (15.5), giving a mention of other methods of measuring neuronal electrical activity in addition to electroencephalography, such as magnetoencephalography and functional magnetic resonance. Conscious that further progress will support a deeper understanding of the temporal course of neuronal activity because of continuous interaction with the environment, we conclude confident that the fractal dimension has begun to uncover important features of the physiology of brain activity and its alterations.

Analysis of EEG features and study of automatic classification in first-episode and drug-naïve patients with major depressive disorder

BACKGROUND

Major depressive disorder (MDD) has a high incidence and an unknown mechanism. There are no objective and sensitive indicators for clinical diagnosis.

OBJECTIVE

This study explored specific electrophysiological indicators and their role in the clinical diagnosis of MDD using machine learning.

METHODS

Forty first-episode and drug-naïve patients with MDD and forty healthy controls (HCs) were recruited. EEG data were collected from all subjects in the resting state with eyes closed for 10 min. The severity of MDD was assessed by the Hamilton Depression Rating Scale (HAMD-17). Machine learning analysis was used to identify the patients with MDD.

RESULTS

Compared to the HC group, the relative power of the low delta and theta bands was significantly higher in the right occipital region, and the relative power of the alpha band in the entire posterior occipital region was significantly lower in the MDD group. In the MDD group, the alpha band scalp functional connectivity was overall lower, while the scalp functional connectivity in the gamma band was significantly higher than that in the HC group. In the feature set of the relative power of the ROI in each band, the highest accuracy of 88.2% was achieved using the KNN classifier while using PCA feature selection. In the explanatory model using SHAP values, the top-ranking influence feature is the relative power of the alpha band in the left parietal region.

CONCLUSIONS

Our findings reveal that the abnormal EEG neural oscillations may reflect an imbalance of excitation, inhibition and hyperactivity in the cerebral cortex in first-episode and drug-naïve patients with MDD. The relative power of the alpha band in the left parietal region is expected to be an objective electrophysiological indicator of MDD.

An Explainable and Robust Deep Learning Approach for Automated Electroencephalography-based Schizophrenia Diagnosis

Schizophrenia (SZ) is a neuropsychiatric disorder that affects millions globally. Current diagnosis of SZ is symptom-based, which poses difficulty due to the variability of symptoms across patients. To this end, many recent studies have developed deep learning methods for automated diagnosis of SZ, especially using raw EEG, which provides high temporal precision. For such methods to be productionized, they must be both explainable and robust. Explainable models are essential to identify biomarkers of SZ, and robust models are critical to learn generalizable patterns, especially amidst changes in the implementation environment. One common example is channel loss during EEG recording, which could be detrimental to classifier performance. In this study, we developed a novel channel dropout (CD) approach to increase the robustness of explainable deep learning models trained on EEG data for SZ diagnosis to channel loss. We developed a baseline convolutional neural network (CNN) architecture and implement our approach as a CD layer added to the baseline (CNN-CD). We then applied two explainability approaches to both models for insight into learned spatial and spectral features and show that the application of CD decreases model sensitivity to channel loss. The CNN and CNN-CD achieved accuracies of 81.9% and 80.9% on testing data, respectively. Furthermore, our models heavily prioritized the parietal electrodes and the α-band, which is supported by existing literature. It is our hope that this study motivates the further development of explainable and robust models and bridges the transition from research to application in a clinical decision support role.

Improving Multichannel Raw Electroencephalography-based Diagnosis of Major Depressive Disorder via Transfer Learning with Single Channel Sleep Stage Data

As the field of deep learning has grown in recent years, its application to the domain of raw resting-state electroencephalography (EEG) has also increased. Relative to traditional machine learning methods or deep learning methods applied to manually engineered features, there are fewer methods for developing deep learning models on small raw EEG datasets. One potential approach for enhancing deep learning performance, in this case, is the use of transfer learning. While a number of studies have presented transfer learning approaches for manually engineered EEG features, relatively few approaches have been developed for raw resting-state EEG. In this study, we propose a novel EEG transfer learning approach wherein we first train a model on a large publicly available single-channel sleep stage classification dataset. We then use the learned representations to develop a classifier for automated major depressive disorder diagnosis with raw multichannel EEG. Statistical testing reveals that our approach significantly improves the performance of our model (p < 0.05), and we also find that the performance of our approach exceeds that of many previous studies using both engineered features and raw EEG. We further examine how transfer learning affected the representations learned by the model through a pair of explainability analyses, identifying key frequency bands and channels utilized across models. Our proposed approach represents a significant step forward for the domain of raw resting-state EEG classification and has broader implications for use with other electrophysiology and time-series modalities. Importantly, it has the potential to expand the use of deep learning methods across a greater variety of raw EEG datasets and lead to the development of more reliable EEG classifiers.

Machine learning-based classification using electroencephalographic multi-paradigms between drug-naïve patients with depression and healthy controls

BACKGROUND

Electroencephalography (EEG) is a supplementary diagnostic tool in psychiatry but lacks practical usage. EEG has demonstrated inconsistent diagnostic ability because major depressive disorder (MDD) is a heterogeneous psychiatric disorder with complex pathologies. In clinical psychiatry, it is essential to detect these complexities using multiple EEG paradigms. Though the application of machine learning to EEG signals in psychiatry has increased, an improvement in its classification performance is still required clinically. We tested the classification performance of multiple EEG paradigms in drug-naïve patients with MDD and healthy controls (HCs).

METHODS

Thirty-one drug-naïve patients with MDD and 31 HCs were recruited in this study. Resting-state EEG (REEG), the loudness dependence of auditory evoked potentials (LDAEP), and P300 were recorded for all participants. Linear discriminant analysis (LDA) and support vector machine (SVM) classifiers with t-test-based feature selection were used to classify patients and HCs.

RESULTS

The highest accuracy was 94.52 % when 14 selected features, including 12 P300 amplitudes (P300A) and two LDAEP features, were layered. The accuracy was 90.32 % when a SVM classifier for 30 selected features (14 P300A, 14 LDAEP, and 2 REEG) was layered in comparison to each REEG, P300A, and LDAEP, the best accuracies of which were 71.57 % (2-layered with LDA), 87.12 % (1-layered with LDA), and 83.87 % (6-layered with SVM), respectively.

LIMITATIONS

The present study was limited by small sample size and difference in formal education year.

CONCLUSIONS

Multiple EEG paradigms are more beneficial than a single EEG paradigm for classifying drug-naïve patients with MDD and HCs.

Automatic depression diagnosis through hybrid EEG and near-infrared spectroscopy features using support vector machine

Depression is a common mental disorder that seriously affects patients' social function and daily life. Its accurate diagnosis remains a big challenge in depression treatment. In this study, we used electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) and measured the whole brain EEG signals and forehead hemodynamic signals from 25 depression patients and 30 healthy subjects during the resting state. On one hand, we explored the EEG brain functional network properties, and found that the clustering coefficient and local efficiency of the delta and theta bands in patients were significantly higher than those in normal subjects. On the other hand, we extracted brain network properties, asymmetry, and brain oxygen entropy as alternative features, used a data-driven automated method to select features, and established a support vector machine model for automatic depression classification. The results showed the classification accuracy was 81.8% when using EEG features alone and increased to 92.7% when using hybrid EEG and fNIRS features. The brain network local efficiency in the delta band, hemispheric asymmetry in the theta band and brain oxygen sample entropy features differed significantly between the two groups (p < 0.05) and showed high depression distinguishing ability indicating that they may be effective biological markers for identifying depression. EEG, fNIRS and machine learning constitute an effective method for classifying depression at the individual level.

Evaluation of Risk of Bias in Neuroimaging-Based Artificial Intelligence Models for Psychiatric Diagnosis: A Systematic Review

IMPORTANCE

Neuroimaging-based artificial intelligence (AI) diagnostic models have proliferated in psychiatry. However, their clinical applicability and reporting quality (ie, feasibility) for clinical practice have not been systematically evaluated.

OBJECTIVE

To systematically assess the risk of bias (ROB) and reporting quality of neuroimaging-based AI models for psychiatric diagnosis.

EVIDENCE REVIEW

PubMed was searched for peer-reviewed, full-length articles published between January 1, 1990, and March 16, 2022. Studies aimed at developing or validating neuroimaging-based AI models for clinical diagnosis of psychiatric disorders were included. Reference lists were further searched for suitable original studies. Data extraction followed the CHARMS (Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modeling Studies) and PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines. A closed-loop cross-sequential design was used for quality control. The PROBAST (Prediction Model Risk of Bias Assessment Tool) and modified CLEAR (Checklist for Evaluation of Image-Based Artificial Intelligence Reports) benchmarks were used to systematically evaluate ROB and reporting quality.

FINDINGS

A total of 517 studies presenting 555 AI models were included and evaluated. Of these models, 461 (83.1%; 95% CI, 80.0%-86.2%) were rated as having a high overall ROB based on the PROBAST. The ROB was particular high in the analysis domain, including inadequate sample size (398 of 555 models [71.7%; 95% CI, 68.0%-75.6%]), poor model performance examination (with 100% of models lacking calibration examination), and lack of handling data complexity (550 of 555 models [99.1%; 95% CI, 98.3%-99.9%]). None of the AI models was perceived to be applicable to clinical practices. Overall reporting completeness (ie, number of reported items/number of total items) for the AI models was 61.2% (95% CI, 60.6%-61.8%), and the completeness was poorest for the technical assessment domain with 39.9% (95% CI, 38.8%-41.1%).

CONCLUSIONS AND RELEVANCE

This systematic review found that the clinical applicability and feasibility of neuroimaging-based AI models for psychiatric diagnosis were challenged by a high ROB and poor reporting quality. Particularly in the analysis domain, ROB in AI diagnostic models should be addressed before clinical application.

Few-Electrode EEG from the Wearable Devices Using Domain Adaptation for Depression Detection

Nowadays, major depressive disorder (MDD) has become a crucial mental disease that endangers human health. Good results have been achieved by electroencephalogram (EEG) signals in the detection of depression. However, EEG signals are time-varying, and the distributions of the different subjects' data are non-uniform, which poses a bad influence on depression detection. In this paper, the deep learning method with domain adaptation is applied to detect depression based on EEG signals. Firstly, the EEG signals are preprocessed and then transformed into pictures by two methods: the first one is to present the three channels of EEG separately in the same image, and the second one is the RGB synthesis of the three channels of EEG. Finally, the training and prediction are performed in the domain adaptation model. The results indicate that the domain adaptation model can effectively extract EEG features and obtain an average accuracy of 77.0 ± 9.7%. This paper proves that the domain adaptation method can effectively weaken the inherent differences of EEG signals, making the diagnosis of different users more accurate.

Altered coordination between frontal delta and parietal alpha networks underlies anhedonia and depressive rumination in major depressive disorder

BACKGROUND

A hyperactive default mode network (DMN) has been observed in people with major depressive disorder (MDD), and weak DMN suppression has been linked to depressive symptoms. However, whether dysregulation of the DMN contributes to blunted positive emotional experience in people with MDD is unclear.

METHODS

We recorded 128-channel electroencephalograms (EEGs) from 24 participants with MDD and 31 healthy controls in a resting state (RS) and an emotion-induction state (ES), in which participants engaged with emotionally positive pictures. We combined Granger causality analysis and data-driven decomposition to extract latent brain networks shared among states and groups, and we further evaluated their interactions across individuals.

RESULTS

We extracted 2 subnetworks. Subnetwork 1 represented a delta (δ)-band (1~4 Hz) frontal network that was activated more in the ES than the RS (i.e., task-positive). Subnetwork 2 represented an alpha (α)-band (8~13 Hz) parietal network that was suppressed more in the ES than the RS (i.e., task-negative). These subnetworks were anticorrelated in both the healthy control and MDD groups, but with different sensitivities: for participants with MDD to achieve the same level of task-positive (subnetwork 1) activation as healthy controls, more suppression of task-negative (subnetwork 2) activation was necessary. Furthermore, the anticorrelation strength in participants with MDD correlated with the severity of 2 core MDD symptoms: anhedonia and rumination.

LIMITATIONS

The sample size was small.

CONCLUSION

Our findings revealed altered coordination between 2 functional networks in MDD and suggest that weak suppression of the task-negative α-band parietal network contributes to blunted positive emotional responses in adults with depression. The subnetworks identified here could be used for diagnosis or targeted for treatment in the future.

Brain-computer interface to predict impulse buying behavior using functional near-infrared spectroscopy

As the rate of vaccination against COVID-19 is increasing, demand for overseas travel is also increasing. Despite people's preference for duty-free shopping, previous studies reported that duty-free shopping increases impulse buying behavior. There are also self-reported tools to measure their impulse buying behavior, but it has the disadvantage of relying on the human memory and perception. Therefore, we propose a Brain-Computer Interface (BCI)-based brain signal processing methodology to supplement these limitations and to reduce ambiguity and conjecture of data. To achieve this goal, we focused on the brain's prefrontal cortex (PFC) activity, which supervises human decision-making and is closely related to impulse buying behavior. The PFC activation is observed by recording signals using a functional near-infrared spectroscopy (fNIRS) while inducing impulse buying behavior in virtual computing environments. We found that impulse buying behaviors were not only higher in online duty-free shops than in online regular stores, but the fNIRS signals were also different on the two sites. We also achieved an average accuracy of 93.78% in detecting impulse buying patterns using a support vector machine. These results were identical to the people's self-reported responses. This study provides evidence as a potential biomarker for detecting impulse buying behavior with fNIRS.

A pilot study of machine learning of resting-state EEG and depression in Parkinson's disease

Introduction

Depression is a non-motor symptom of Parkinson's disease (PD). PD-related depression is difficult to diagnose, and the neurophysiological basis is poorly understood. Depression can markedly affect cortical function, which suggests that scalp electroencephalography (EEG) may be able to distinguish depression in PD. We conducted a pilot study of depression and resting-state EEG in PD.

Methods

We recruited 18 PD patients without depression, 18 PD patients with depression, and 12 demographically similar non-PD patients with clinical depression. All patients were on their usual medications. We collected resting-state EEG in all patients and compared cortical brain signal features between patients with and without depression. We used a machine learning algorithm that harnesses the entire power spectrum (linear predictive coding of EEG Algorithm for PD: LEAPD) to distinguish between groups.

Results

We found differences between PD patients with and without depression in the alpha band (8-13 Hz) globally and in the beta (13-30 Hz) and gamma (30-50 Hz) bands in the central electrodes. From two minutes of resting-state EEG, we found that LEAPD-based machine learning could robustly distinguish between PD patients with and without depression with 97 % accuracy and between PD patients with depression and non-PD patients with depression with 100 % accuracy. We verified the robustness of our finding by confirming that the classification accuracy gracefully declines as data are randomly truncated.

Conclusions

Our results suggest that resting-state EEG power spectral analysis has the potential to distinguish depression in PD accurately. We demonstrated the efficacy of the LEAPD algorithm in identifying PD patients with depression from PD patients without depression and controls with depression. Our data provide insight into cortical mechanisms of depression and could lead to novel neurophysiological markers for non-motor symptoms of PD.

Mental State Classification Using Multi-Graph Features

We consider the problem of extracting features from passive, multi-channel electroencephalogram (EEG) devices for downstream inference tasks related to high-level mental states such as stress and cognitive load. Our proposed feature extraction method uses recently developed spectral-based multi-graph tools and applies them to the time series of graphs implied by the statistical dependence structure (e.g., correlation) amongst the multiple sensors. We study the features in the context of two datasets each consisting of at least 30 participants and recorded using multi-channel EEG systems. We compare the classification performance of a classifier trained on the proposed features to a classifier trained on the traditional band power-based features in three settings and find that the two feature sets offer complementary predictive information. We conclude by showing that the importance of particular channels and pairs of channels for classification when using the proposed features is neuroscientifically valid.

Smart Electronic Textiles for Wearable Sensing and Display

Increasing demand of using everyday clothing in wearable sensing and display has synergistically advanced the field of electronic textiles, or e-textiles. A variety of types of e-textiles have been formed into stretchy fabrics in a manner that can maintain their intrinsic properties of stretchability, breathability, and wearability to fit comfortably across different sizes and shapes of the human body. These unique features have been leveraged to ensure accuracy in capturing physical, chemical, and electrophysiological signals from the skin under ambulatory conditions, while also displaying the sensing data or other immediate information in daily life. Here, we review the emerging trends and recent advances in e-textiles in wearable sensing and display, with a focus on their materials, constructions, and implementations. We also describe perspectives on the remaining challenges of e-textiles to guide future research directions toward wider adoption in practice.

Changes of adrenocorticotropic hormone rhythm and cortisol circadian rhythm in patients with depression complicated with anxiety and their effects on the psychological state of patients

Objective: This work was to explore the rhythm of adrenocorticotropic hormone (ACTH) and cortisol in patients with depression and anxiety and their effects on mental state. In this work, with depression complicated with anxiety patients as the A-MDD group (n = 21), and depression without anxiety symptoms as the NA-MDD group (n = 21). Firstly, data features were extracted according to the electroencephalo-graph (EEG) data of different patients, and a DR model was constructed for diagnosis. The Hamilton Depression Scale 24 (HAMD-24) was employed to evaluate the severity, and the ACTH and cortisol levels were detected and compared for patients in the A-MDD group and NA-MDD group. In addition, the psychological status of the patients was assessed using the Toronto Alexithymia Scale (TAS). As a result, the AI-based DR model showed a high recognition accuracy for depression. The HAMD-24 score in the A-MDD group (31.81 ± 5.39 points) was statistically higher than the score in the NA-MDD group (25.25 ± 5.02 points) (P < 0.05). No visible difference was found in ACTH levels of patients in different groups (P > 0.05). The incidence of cortisol rhythm disorder (CRD) in the A-MDD group was much higher (P < 0.05). The differences in TAS scores between the two groups were significantly statistically significant (P < 0.01). In conclusion, the AI-based DR Model achieves a more accurate identification of depression; depression with or without anxiety has different effects on the mental state of patients. CRD may be one of the biological markers of depression combined with anxiety.

Machine learning approaches for diagnosing depression using EEG: A review

Depression has become one of the most crucial public health issues, threatening the quality of life of over 300 million people throughout the world. Nevertheless, the clinical diagnosis of depression is now still hampered by behavioral diagnostic methods. Due to the lack of objective laboratory diagnostic criteria, accurate identification and diagnosis of depression remained elusive. With the rise of computational psychiatry, a growing number of studies have combined resting-state electroencephalography with machine learning (ML) to alleviate diagnosis of depression in recent years. Despite the exciting results, these were worrisome of these studies. As a result, ML prediction models should be continuously improved to better screen and diagnose depression. Finally, this technique would be used for the diagnosis of other psychiatric disorders in the future.