Depression diagnosis by deep learning using EEG signals: A systematic review

Determining human-coronavirus protein-protein interaction using machine intelligence

“…Following a comprehensive literature review, the authors of [ 92 ] described the depression diagnosis of the human brain from electroencephalogram EEG signals using various deep learning methods. Therefore, in the neuro analysis of the human brain, machine learning, and deep learning algorithms are efficacious, which further attests to the significance of the proposed ensemble learning approach.…”

Utility of Cognitive Neural Features for Predicting Mental Health Behaviors

“…In this study, we performed logistic regression-based prediction analyses of neural spectral activations that were simultaneously recorded while adult participants engaged in a suite of cognitive task assessments, in order to predict mental health symptoms. Though many deep neural networks are gaining momentum for understanding biomarkers of psychiatric disorders [15][16][17], there is a school of thought that favors the development of interpretable machine learning models [63][64][65][66], instead of deep learning black-box models, to gain insights into clinical translation. In our study, with the help of simplistic regression methods and guided feature augmentation, we propose that our findings are more interpretable and traceable to the neurophysiological and neuroimaging literature and that our findings are more amenable to translation into clinical models.…”

“…Many studies suggest that the EEG biomarkers for mental health disorders, such as anxiety [9][10][11], depression [12], and inattentive ADHD [13,14]. Some deep neural networks applied to electroencephalography data claim greater than 90% accuracy in terms of predicting mental health symptoms [15][16][17]. Also, traditionally, several studies relate EEG neural markers to cognitive functions, such as attention [18][19][20], inhibitory control [21][22][23], emotion processing [18,24], working memory, and cognitive load [25][26][27][28].…”

Machine Learning in Neuroimaging for Diagnosis of Major Depressive Disorder

“…For example, after building EEG-based diagnosis model using through deep convolutional neural network (CNN) for classifying MDD patients from healthy controls, Uyulan, C. et al ( 2020) was able to reveal that the higher average delta amplitude in MDD than that of healthy control subjects may indicate a biomarker for MDD, and proposed that the methodology can be adapted to computer-assisted diagnosis of MDD to validate clinical diagnosis [32]. However, Safayari, A., & Bolhasani, H. (2021) regarded that these models could pose some implementation challenges to clinical environments due to complexity and novelty [33]. Among similar studies, the lack of sufficient data to improve or evaluate the accuracy of the DL-based methods is one the main barriers to clinical utility [33].…”

“…As a consequence, machine learning approaches are integrated into neuroimaging to provide multivariate solutions that demonstrate greater sensitivity and more reliable predictions than univariate methods, thus enabling the development of imaging brain signatures at the individual level [17,18]. Among the most successful machine learning techniques, support vector machines (SVMs) [21][22][23], random forests (RF) [24][25][26][27][28][29][30], and deep learning (DL) [31][32][33] have become exceedingly popular for neuroimaging analysis in recent years.…”