An Improved Classification Model for Depression Detection Using EEG and Eye Tracking Data

Depression Detection and Diagnosis Based on Electroencephalogram (EEG) Analysis: A Systematic Review

Background: Mental disorders are disturbances of brain functions that cause cognitive, affective, volitional, and behavioral functions to be disrupted to varying degrees. One of these disorders is depression, a significant factor contributing to the increase in suicide cases worldwide. Consequently, depression has become a significant public health issue globally. Electroencephalogram (EEG) data can be utilized to diagnose mild depression disorder (MDD), offering valuable insights into the pathophysiological mechanisms underlying mental disorders and enhancing the understanding of MDD. Methods: This survey emphasizes the critical role of EEG in advancing artificial intelligence (AI)-driven approaches for depression diagnosis. By focusing on studies that integrate EEG with machine learning (ML) and deep learning (DL) techniques, we systematically analyze methods utilizing EEG signals to identify depression biomarkers. The survey highlights advancements in EEG preprocessing, feature extraction, and model development, showcasing how these approaches enhance the diagnostic precision, scalability, and automation of depression detection. Results: This survey is distinguished from prior reviews by addressing their limitations and providing researchers with valuable insights for future studies. It offers a comprehensive comparison of ML and DL approaches utilizing EEG and an overview of the five key steps in depression detection. The survey also presents existing datasets for depression diagnosis and critically analyzes their limitations. Furthermore, it explores future directions and challenges, such as enhancing diagnostic robustness with data augmentation techniques and optimizing EEG channel selection for improved accuracy. The potential of transfer learning and encoder-decoder architectures to leverage pre-trained models and enhance diagnostic performance is also discussed. Advancements in feature extraction methods for automated depression diagnosis are highlighted as avenues for improving ML and DL model performance. Additionally, integrating Internet of Things (IoT) devices with EEG for continuous mental health monitoring and distinguishing between different types of depression are identified as critical research areas. Finally, the review emphasizes improving the reliability and predictability of computational intelligence-based models to advance depression diagnosis. Conclusions: This study will serve as a well-organized and helpful reference for researchers working on detecting depression using EEG signals and provide insights into the future directions outlined above, guiding further advancements in the field.

Opportunities and Challenges for Clinical Practice in Detecting Depression Using EEG and Machine Learning

Major depressive disorder (MDD) is associated with substantial morbidity and mortality, yet its diagnosis and treatment rates remain low due to its diverse and often overlapping clinical manifestations. In this context, electroencephalography (EEG) has gained attention as a potential objective tool for diagnosing depression. This study aimed to evaluate the effectiveness of EEG in identifying MDD by analyzing 140 EEG recordings from patients diagnosed with depression and healthy volunteers. Using various machine learning (ML) classification models, we achieved up to 80% accuracy in distinguishing individuals with MDD from healthy controls. Despite its promise, this approach has limitations. The variability in the clinical and biological presentations of depression, as well as patient-specific confounding factors, must be carefully considered when integrating ML technologies into clinical practice. Nevertheless, our findings suggest that an EEG-based ML model holds potential as a diagnostic aid for MDD, paving the way for further refinement and clinical application.

Synchronization-based fusion of EEG and eye blink signals for enhanced decoding accuracy

Decoding locomotor tasks is crucial in cognitive neuroscience for understanding brain responses to physical tasks. Traditional methods like EEG offer brain activity insights but may require additional modalities for enhanced interpretative precision and depth. The integration of EEG with ocular metrics, particularly eye blinks, presents a promising avenue for understanding cognitive processes by combining neural and ocular behaviors. However, synchronizing EEG and eye blink activities poses a significant challenge due to their frequently inconsistent alignment. Our study with 35 participants performing various locomotor tasks such as standing, walking, and transversing obstacles introduced a novel methodology, pcEEG+, which fuses EEG principal components (pcEEG) with aligned eye blink data (syncBlink). The results demonstrated that pcEEG+ significantly improved decoding accuracy in locomotor tasks, reaching 78% in some conditions, and surpassed standalone pcEEG and syncBlink methods by 7.6% and 22.7%, respectively. The temporal generalization matrix confirmed the consistency of pcEEG+ across tasks and times. The results were replicated using two driving simulator datasets, thereby confirming the validity of our method. This study demonstrates the efficacy of the pcEEG+ method in decoding locomotor tasks, underscoring the importance of temporal synchronization for accuracy and offering a deeper insight into brain activity during complex movements.

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.

MDD brain network analysis based on EEG functional connectivity and graph theory

Background

Existing studies have shown that the brain network of major depression disorder (MDD) has abnormal topologies. However, constructing reliable MDD brain networks is still an open problem.

New method

This paper proposed a reliable MDD brain network construction method. First, seven connectivity methods are used to calculate the correlation between channels and obtain the functional connectivity matrix. Then, the matrix is binarized using four binarization methods to obtain the EEG brain network. Besides, we proposed an improved binarization method based on the criterion of maximizing differences between groups: the adaptive threshold (AT) method. The AT can automatically set the optimal binarization threshold and overcome the artificial influence of traditional methods. After that, several network metrics are extracted from the brain network to analyze inter-group differences. Finally, we used statistical analysis and Fscore values to compare the performance of different methods and establish the most reliable method for brain network construction.

Results

In theta, alpha, and total frequency bands, the clustering coefficient, global efficiency, local efficiency, and degree of the MDD brain network decrease, and the path length of the MDD brain network increases.

Comparison with existing methods

The results show that AT outperforms the existing binarization methods. Compared with other methods, the brain network construction method based on phase-locked value (PLV) and AT has better reliability.

Conclusions

MDD has brain dysfunction, particularly in the frontal and temporal lobes.

Diagnosing and tracking depression based on eye movement in response to virtual reality

Introduction

Depression is a prevalent mental illness that is primarily diagnosed using psychological and behavioral assessments. However, these assessments lack objective and quantitative indices, making rapid and objective detection challenging. In this study, we propose a novel method for depression detection based on eye movement data captured in response to virtual reality (VR).

Methods

Eye movement data was collected and used to establish high-performance classification and prediction models. Four machine learning algorithms, namely eXtreme Gradient Boosting (XGBoost), multilayer perceptron (MLP), Support Vector Machine (SVM), and Random Forest, were employed. The models were evaluated using five-fold cross-validation, and performance metrics including accuracy, precision, recall, area under the curve (AUC), and F1-score were assessed. The predicted error for the Patient Health Questionnaire-9 (PHQ-9) score was also determined.

Results

The XGBoost model achieved a mean accuracy of 76%, precision of 94%, recall of 73%, and AUC of 82%, with an F1-score of 78%. The MLP model achieved a classification accuracy of 86%, precision of 96%, recall of 91%, and AUC of 86%, with an F1-score of 92%. The predicted error for the PHQ-9 score ranged from -0.6 to 0.6.To investigate the role of computerized cognitive behavioral therapy (CCBT) in treating depression, participants were divided into intervention and control groups. The intervention group received CCBT, while the control group received no treatment. After five CCBT sessions, significant changes were observed in the eye movement indices of fixation and saccade, as well as in the PHQ-9 scores. These two indices played significant roles in the predictive model, indicating their potential as biomarkers for detecting depression symptoms.

Discussion

The results suggest that eye movement indices obtained using a VR eye tracker can serve as useful biomarkers for detecting depression symptoms. Specifically, the fixation and saccade indices showed promise in predicting depression. Furthermore, CCBT demonstrated effectiveness in treating depression, as evidenced by the observed changes in eye movement indices and PHQ-9 scores. In conclusion, this study presents a novel approach for depression detection using eye movement data captured in VR. The findings highlight the potential of eye movement indices as biomarkers and underscore the effectiveness of CCBT in treating depression.

Research on a New Intelligent and Rapid Screening Method for Depression Risk in Young People Based on Eye Tracking Technology

Depression is a prevalent mental disorder, with young people being particularly vulnerable to it. Therefore, we propose a new intelligent and rapid screening method for depression risk in young people based on eye tracking technology. We hypothesized that the "emotional perception of eye movement" could characterize defects in emotional perception, recognition, processing, and regulation in young people at high risk for depression. Based on this hypothesis, we designed the "eye movement emotional perception evaluation paradigm" and extracted digital biomarkers that could objectively and accurately evaluate "facial feature perception" and "facial emotional perception" characteristics of young people at high risk of depression. Using stepwise regression analysis, we identified seven digital biomarkers that could characterize emotional perception, recognition, processing, and regulation deficiencies in young people at high risk for depression. The combined effectiveness of an early warning can reach 0.974. Our proposed technique for rapid screening has significant advantages, including high speed, high early warning efficiency, low cost, and high intelligence. This new method provides a new approach to help effectively screen high-risk individuals for depression.

Machine Learning and Electroencephalogram Signal based Diagnosis of Dipression

Depression is a psychological condition which hampers day to day activity (Thinking, Feeling or Action). The early detection of this illness will help to save many lives because it is now recognized as a global problem which could even lead to suicide. Electroencephalogram (EEG) signals can be used to diagnose depression using machine learning techniques. The dataset studied is public dataset which consists of 30 healthy people and 34 depression patients. The methods used for detection of depression are Decision Tree, Random Forest, Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Bidirectional Long-Short Term Memory (Bi-LSTM), Gradient Boosting, Extreme Gradient Boosting (XGBoost) along with band power. Among Deep Learning techniques, CNN model got the highest accuracy with 98.13%, specificity of 99%, and sensitivity of 97% using band power features.

Depression and Cognitive Impairment: Current Understanding of Its Neurobiology and Diagnosis

BACKGROUND

Eye movement is critical for obtaining precise visual information and providing sensorimotor processes and advanced cognitive functions to the brain behavioral indicator.

METHODS

In this article, we present a narrative review of the eye-movement paradigms (such as fixation, smooth pursuit eye movements, and memory-guided saccade tasks) in major depression.

RESULTS

Characteristics of eye movement are considered to reflect several aspects of cognitive deficits regarded as an aid to diagnosis. Findings regarding depressive disorders showed differences from the healthy population in paradigms, the characteristics of eye movement may reflect cognitive deficits in depression. Neuroimaging studies have demonstrated the effectiveness of different eye movement paradigms for MDD screening.

CONCLUSION

Depression can be distinguished from other mental illnesses based on eye movements. Eye movement reflects cognitive deficits that can help diagnose depression, and it can make the entire diagnostic process more accurate.

Content-based multiple evidence fusion on EEG and eye movements for mild depression recognition

BACKGROUND AND OBJECTIVE

Depression is a serious neurological disorder that has become a major health problem worldwide. The detection of mild depression is important for the diagnosis of depression in early stages. This research seeks to find a more accurate fusion model which can be used for mild depression detection using Electroencephalography and eye movement data.

METHODS

This study proposes a content-based multiple evidence fusion (CBMEF) method, which fuses EEG and eye movement data at decision level. The method mainly includes two modules, the classification performance matrix module and the dual-weight fusion module. The classification performance matrices of different modalities are estimated by Bayesian rule based on confusion matrix and Mahalanobis distance, and the matrices were used to correct the classification results. Then the relative conflict degree of each modality is calculated, and different weights are assigned to the above modalities at the decision fusion layer according to this conflict degree.

RESULTS

The experimental results show that the proposed method outperforms other fusion methods as well as the single modality results. The highest accuracies achieved 91.12%, and sensitivity, specificity and precision were 89.20%, 93.03%, 92.76%.

CONCLUSIONS

The promising results showed the potential of the proposed approach for the detection of mild depression. The idea of introducing the classification performance matrix and the dual-weight model to multimodal biosignals fusion casts a new light on the researches of depression recognition.

EEG diagnosis of depression based on multi-channel data fusion and clipping augmentation and convolutional neural network

Depression is an undetectable mental disease. Most of the patients with depressive symptoms do not know that they are suffering from depression. Since the novel Coronavirus pandemic 2019, the number of patients with depression has increased rapidly. There are two kinds of traditional depression diagnosis. One is that professional psychiatrists make diagnosis results for patients, but it is not conducive to large-scale depression detection. Another is to use electroencephalography (EEG) to record neuronal activity. Then, the features of the EEG are extracted using manual or traditional machine learning methods to diagnose the state and type of depression. Although this method achieves good results, it does not fully utilize the multi-channel information of EEG. Aiming at this problem, an EEG diagnosis method for depression based on multi-channel data fusion cropping enhancement and convolutional neural network is proposed. First, the multi-channel EEG data are transformed into 2D images after multi-channel fusion (MCF) and multi-scale clipping (MSC) augmentation. Second, it is trained by a multi-channel convolutional neural network (MCNN). Finally, the trained model is loaded into the detection device to classify the input EEG signals. The experimental results show that the combination of MCF and MSC can make full use of the information contained in the single sensor records, and significantly improve the classification accuracy and clustering effect of depression diagnosis. The method has the advantages of low complexity and good robustness in signal processing and feature extraction, which is beneficial to the wide application of detection systems.

Perception Exploration on Robustness Syndromes With Pre-processing Entities Using Machine Learning Algorithm

The majority of the current-generation individuals all around the world are dealing with a variety of health-related issues. The most common cause of health problems has been found as depression, which is caused by intellectual difficulties. However, most people are unable to recognize such occurrences in them, and no procedures for discriminating them from normal people have been created so far. Even some advanced technologies do not support distinct classes of individuals as language writing skills vary greatly across numerous places, making the central operations cumbersome. As a result, the primary goal of the proposed research is to create a unique model that can detect a variety of diseases in humans, thereby averting a high level of depression. A machine learning method known as the Convolutional Neural Network (CNN) model has been included into this evolutionary process for extracting numerous features in three distinct units. The CNN also detects early-stage problems since it accepts input in the form of writing and sketching, both of which are turned to images. Furthermore, with this sort of image emotion analysis, ordinary reactions may be easily differentiated, resulting in more accurate prediction results. The characteristics such as reference line, tilt, length, edge, constraint, alignment, separation, and sectors are analyzed to test the usefulness of CNN for recognizing abnormalities, and the extracted features provide an enhanced value of around 74%higher than the conventional models.

Effective differentiation between depressed patients and controls using discriminative eye movement features

BACKGROUND

Depression is a common debilitating mental disorder caused by various factors. Identifying and diagnosing depression are challenging because the clinical evaluation of depression is mainly subjective, lacking objective and quantitative indicators. The present study investigated the value and significance of eye movement measurements in distinguishing depressed patients from controls.

METHODS

Ninety-five depressed patients and sixty-nine healthy controls performed three eye movement tests, including fixation stability, free-viewing, and anti-saccade tests, and eleven eye movement indexes were obtained from these tests. The independent t-test was adopted for group comparisons, and multiple logistic regression analysis was employed to identify diagnostic biomarkers. Support vector machine (SVM), quadratic discriminant analysis (QDA), and Bayesian (BYS) algorithms were applied to build the classification models.

RESULTS

Depressed patients exhibited eye movement anomalies, characterized by increased saccade amplitude in the fixation stability test; diminished saccade velocity in the anti-saccade test; and reduced saccade amplitude, shorter scan path length, lower saccade velocity, decreased dynamic range of pupil size, and lower pupil size ratio in the free-viewing test. Four features mentioned above entered the logistic regression equation. The classification accuracies of SVM, QDA, and BYS models reached 86.0%, 81.1%, and 83.5%, respectively.

CONCLUSIONS

Depressed patients exhibited abnormalities across multiple tests of eye movements, assisting in differentiating depressed patients from healthy controls in a cost-effective and non-invasive manner.

Multimodal EEG and Eye Tracking Feature Fusion Approaches for Attention Classification in Hybrid BCIs

Often, various modalities capture distinct aspects of particular mental states or activities. While machine learning algorithms can reliably predict numerous aspects of human cognition and behavior using a single modality, they can benefit from the combination of multiple modalities. This is why hybrid BCIs are gaining popularity. However, it is not always straightforward to combine features from a multimodal dataset. Along with the method for generating the features, one must decide when the modalities should be combined during the classification process. We compare unimodal EEG and eye tracking classification of internally and externally directed attention to multimodal approaches for early, middle, and late fusion in this study. On a binary dataset with a chance level of 0.5, late fusion of the data achieves the highest classification accuracy of 0.609–0.675 (95%-confidence interval). In general, the results indicate that for these modalities, middle or late fusion approaches are better suited than early fusion approaches. Additional validation of the observed trend will require the use of additional datasets, alternative feature generation mechanisms, decision rules, and neural network designs. We conclude with a set of premises that need to be considered when deciding on a multimodal attentional state classification approach.

Predicting the Reader's English Level from Reading Fixation Patterns Using the Siamese Convolutional Neural Network

Abstract-Among numerous functions performed by the human eye, reading is a common task that best reflects an individual's understanding and cognitive patterns. Previous studies showed that text comprehension may be determined by comprehension monitoring, a metacognitive process that evaluates and regulates the pattern of comprehension. Herein, we propose a hypothesis: an individual's cognitive pattern during reading is predictive of the level of reading comprehension. According to the criteria of the College English Test Band Six (CET-6), 80 participants (sophomore and junior) were divided into a pass group (n = 40) and a non-pass group (n = 40). Heatmaps of eye fixation counts were collected by an eye-tracker while each participant executed four reading comprehension tests. Using these heatmaps as inputs, we proposed the Siamese convolutional neural network models to predict the English level of participants. Both strategies of "Trained from scratch" and "Pretrained with fine-tuning" were employed. "Soft Voting" was applied to integrate the predictions from four tests. Results showed that the Siamese network model trained by the datasets with the cluster radius of fixation equal to 25 pixels and connection layer by L1 norm distance has a satisfactory or superior performance to other comparative experiments. The AUC values of Siamese networks trained by the two strategies reach 0.941 and 0.956, respectively. This indicates that the individual reading cognitive pattern captured by the eye-tracker can predict the level of reading comprehension through advanced deep learning models.

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.

Affective Computing for Late-Life Mood and Cognitive Disorders

Affective computing (also referred to as artificial emotion intelligence or emotion AI) is the study and development of systems and devices that can recognize, interpret, process, and simulate emotion or other affective phenomena. With the rapid growth in the aging population around the world, affective computing has immense potential to benefit the treatment and care of late-life mood and cognitive disorders. For late-life depression, affective computing ranging from vocal biomarkers to facial expressions to social media behavioral analysis can be used to address inadequacies of current screening and diagnostic approaches, mitigate loneliness and isolation, provide more personalized treatment approaches, and detect risk of suicide. Similarly, for Alzheimer's disease, eye movement analysis, vocal biomarkers, and driving and behavior can provide objective biomarkers for early identification and monitoring, allow more comprehensive understanding of daily life and disease fluctuations, and facilitate an understanding of behavioral and psychological symptoms such as agitation. To optimize the utility of affective computing while mitigating potential risks and ensure responsible development, ethical development of affective computing applications for late-life mood and cognitive disorders is needed.

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

Major depressive disorder (MDD) is a global healthcare issue and one of the leading causes of disability. Machine learning combined with non-invasive electroencephalography (EEG) has recently been shown to have the potential to diagnose MDD. However, most of these studies analyzed small samples of participants recruited from a single source, raising serious concerns about the generalizability of these results in clinical practice. Thus, it has become critical to re-evaluate the efficacy of various common EEG features for MDD detection across large and diverse datasets. To address this issue, we collected resting-state EEG data from 400 participants across four medical centers and tested classification performance of four common EEG features: band power (BP), coherence, Higuchi's fractal dimension, and Katz's fractal dimension. Then, a sequential backward selection (SBS) method was used to determine the optimal subset. To overcome the large data variability due to an increased data size and multi-site EEG recordings, we introduced the conformal kernel (CK) transformation to further improve the MDD as compared with the healthy control (HC) classification performance of support vector machine (SVM). The results show that (1) coherence features account for 98% of the optimal feature subset; (2) the CK-SVM outperforms other classifiers such as K-nearest neighbors (K-NN), linear discriminant analysis (LDA), and SVM; (3) the combination of the optimal feature subset and CK-SVM achieves a high five-fold cross-validation accuracy of 91.07% on the training set (140 MDD and 140 HC) and 84.16% on the independent test set (60 MDD and 60 HC). The current results suggest that the coherence-based connectivity is a more reliable feature for achieving high and generalizable MDD detection performance in real-life clinical practice.

Characterization of specific spatial functional connectivity difference in depression during sleep

Depression is a common mental illness and a large number of researchers have been still devoted to exploring effective biomarkers for the identification of depression. Few researches have been conducted on functional connectivity (FC) during sleep in depression. In this paper, a novel depression characterization is proposed using specific spatial FC features of sleep electroencephalography (EEG). Overnight polysomnography recordings were obtained from 26 healthy individuals and 25 patients with depression. The weighted phase lag indexes (WPLIs) of four frequency bands and five sleep periods were obtained from 16 EEG channels. The high discriminative connections extracted via feature evaluation and the cross-within variation (CW)-the spatial feature constructed to characterize the different performances in inter- and intra-hemispheric FC based on WPLIs, were utilized to classify patients and normal controls. The results showed that enhanced average FC and spatial differences, higher inter-hemispheric FC and lower intra-hemispheric FC, were found in patients. Furthermore, abnormalities in the inter-hemispheric connections of the temporal lobe in the theta band should be important indicators of depression. Finally, both CW and high discriminative WPLI features performed well in depression screening and CW was more specific for characterizing abnormal cortical EEG performance of depression. Our work investigated and characterized the abnormalities in sleep cortical activity in patients with depression, and may provide potential biomarkers for assisting with depression identification and new insights into the understanding of pathological mechanisms in depression.

Business Simulation Games Analysis Supported by Human-Computer Interfaces: A Systematic Review

This article performs a Systematic Review of studies to answer the question: What are the researches related to the learning process with (Serious) Business Games using data collection techniques with Electroencephalogram or Eye tracking signals? The PRISMA declaration method was used to guide the search and inclusion of works related to the elaboration of this study. The 19 references resulting from the critical evaluation initially point to a gap in investigations into using these devices to monitor serious games for learning in organizational environments. An approximation with equivalent sensing studies in serious games for the contribution of skills and competencies indicates that continuous monitoring measures, such as mental state and eye fixation, proved to identify the players' attention levels effectively. Also, these studies showed effectiveness in the flow at different moments of the task, motivating and justifying the replication of these studies as a source of insights for the optimized design of business learning tools. This study is the first systematic review and consolidates the existing literature on user experience analysis of business simulation games supported by human-computer interfaces.

Interpreting Depression From Question-wise Long-term Video Recording of SDS Evaluation

Self-Rating Depression Scale (SDS) questionnaire has frequently been used for efficient depression preliminary screening. However, the uncontrollable self-administered measure can be easily affected by insouciantly or deceptively answering, and producing the different results with the clinician-administered Hamilton Depression Rating Scale (HDRS) and the final diagnosis. Clinically, facial expression (FE) and actions play a vital role in clinician-administered evaluation, while FE and action are underexplored for self-administered evaluations. In this work, we collect a novel dataset of 200 subjects to evidence the validity of self-rating questionnaires with their corresponding question-wise video recording. To automatically interpret depression from the SDS evaluation and the paired video, we propose an end-to-end hierarchical framework for the long-term variable-length video, which is also conditioned on the questionnaire results and the answering time. Specifically, we resort to a hierarchical model which utilizes a 3D CNN for local temporal pattern exploration and a redundancy-aware self-attention (RAS) scheme for question-wise global feature aggregation. Targeting for the redundant long-term FE video processing, our RAS is able to effectively exploit the correlations of each video clip within a question set to emphasize the discriminative information and eliminate the redundancy based on feature pair-wise affinity. Then, the question-wise video feature is concatenated with the questionnaire scores for final depression detection. Our thorough evaluations also show the validity of fusing SDS evaluation and its video recording, and the superiority of our framework to the conventional state-of-the-art temporal modeling methods.