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.

A fractal scaling analysis of the SARS-CoV-2 genome sequence

An approach based on fractal scaling analysis to characterize the organization of the SARS-CoV-2 genome sequence was used. The method is based on the detrended fluctuation analysis (DFA) implemented on a sliding window scheme to detect variations of long-range correlations over the genome sequence regions. The nucleotides sequence is mapped in a numerical sequence by using four different assignation rules: amino-keto, purine-pyrimidine, hydrogen-bond and hydrophobicity patterns. The originally reported sequence from Wuhan isolates (Wuhan Hu-1) was considered as a reference to contrast the structure of the 2002-2004 SARS-CoV-1 strain. Long-range correlations, quantified in terms of a scaling exponent, depended on both the mapping rule and the sequence region. Deviations from randomness were attributed to serial correlations or anti-correlations, which can be ascribed to ordered regions of the genome sequence. It was found that the Wuhan Hu-1 sequence was more random than the SARS-CoV-1 sequence, which suggests that the SARS-CoV-2 possesses a more efficient genomic structure for replication and infection. In general, the virus isolated in the early 2020 months showed slight correlation differences with the Wuhan Hu-1 sequence. However, early isolates from India and Italy presented visible differences that led to a more ordered sequence organization. It is apparent that the increased sequence order, particularly in the spike region, endowed some early variants with a more efficient mechanism to spreading, replicating and infecting. Overall, the results showed that the DFA provides a suitable framework to assess long-term correlations hidden in the internal organization of the SARS-CoV-2 genome sequence.

Detection of Depression and Scaling of Severity Using Six Channel EEG Data

Depression is a psychiatric problem which affects the growth of a person, like how a person thinks, feels and behaves. The major reason behind wrong diagnosis of depression is absence of any laboratory test for detection as well as severity scaling of depression. Any degradation in the working of the brain can be identified through change in the electroencephalogram (EEG) signal. Thus detection as well as severity scaling of depression is done in this study using EEG signal. In this study, features are extracted from the temporal region of the brain using six (FT7, FT8, T7, T8, TP7, TP8) channels. The linear features used are delta, theta, alpha, beta, gamma1 and gamma2 band power and their corresponding asymmetry as well as paired asymmetry. The non-linear features used are Sample Entropy (SampEn) and Detrended Fluctuation Analysis (DFA). The classifiers used are: Bagging along with three different kernel functions (Polynomial, Gaussian and Sigmoidal) of Support Vector Machine (SVM). Feature selection technique used is ReliefF. Highest classification accuracy of 96.02% and 79.19% was achieved for detection and severity scaling of depression using SVM (Gaussian Kernel Function) and ReliefF as feature selection. From the analysis, it was found that depression affects the temporal region of the brain (temporo-parietal region).It was also found that depression affects the higher frequency band features more and it affects each hemisphere differently. It can also be analysed that out of all the kernel of SVM, Gaussian kernel is more efficient to other kernels. Of all the features, combination of all paired asymmetry and asymmetry showed high classification accuracy (accuracy of 90.26% for detection of depression and accuracy of 75.31% for severity scaling).