Symbolic Transfer Entropy Research Articles

Using deep learning and pretreatment EEG to predict response to sertraline, bupropion, and placebo

ObjectivePredicting an individual’s response to antidepressant medication remains one of the most challenging tasks in the treatment of major depressive disorder (MDD). Our objective was to use the large EMBARC study database to develop an electroencephalography (EEG)-based method to predict response to antidepressant treatment. MethodsPre-treatment EEG data were collected from study participants treated with either sertraline (N = 105), placebo (N = 119), or bupropion (N = 35). After preprocessing, the robust exact low-resolution electromagnetic tomography (ReLORETA) brain source localization method was used to reconstruct the source signals in 54 brain regions. Connectivity between regions was determined using symbolic transfer entropy (STE). A convolutional neural network (CNN) classified participants as responders or non-responders to each treatment. ResultsClassification accuracy was 91.0%, 95.4%, and 86.8% for sertraline, placebo, and bupropion, respectively. The most highly predictive features were connectivity between i) the anterior cingulate cortex and superior parietal lobule (alpha frequency), ii) the anterior cingulate cortex and orbitofrontal area (beta frequency), and iii) the orbitofrontal area and anterior cingulate cortex (gamma frequency). ConclusionCNN analysis of EEG connectivity may accurately predict response to sertraline, bupropion, and placebo. SignificanceThe suggested method may offer clinicians an accessible and cost-effective tool for speedy treatment and helps pharmaceutical firms to test new antidepressants efficiently.

Feedback directions governing self-sustained thermoacoustic instability in rocket engine combustors

The occurrence of high-frequency (>1000 Hz) thermoacoustic instability (TAI) sustained by mutual feedback among the acoustic field, heat release rate oscillations, and hydrodynamic oscillations poses severe challenges to the operation and structural integrity of rocket engines. Hence, quantifying the differing levels of feedback between these variables can help uncover the underlying mechanisms behind such high-frequency TAI, enabling redesign of combustors to mitigate TAI. However, so far, no concrete method exists to decipher the varying levels of mutual feedback during high-frequency TAI. In the present study, we holistically investigate the mutual influence based on the spatiotemporal directionality among acoustic pressure, heat release rate, and hydrodynamic and thermal oscillations during TAI of a single-element rocket engine combustor. Using symbolic transfer entropy, we identify the spatiotemporal direction of feedback interactions between those primary variables when acoustic waves significantly emerge during TAI. We unveil the influence of vorticity dynamics at the fuel collar (or the propellant splitter plate) as the primary stimulant over the heat release rate fluctuations to rapidly amplify the amplitude of the acoustic field. Furthermore, depending on the quantification of the degree of the mutual information (i.e., the net direction of information), we identify the switches in dominating the thermoacoustic driving between the variables during TAI, each representing a distinct mechanism of a thermoacoustic state. Additionally, from this quantification, we analyze the relative dominance of the variables and rank-order the mutual feedback according to their impact on driving TAI.

A novel domain adaptation method with physical constraints for shale gas production forecasting

Effective forecasting of shale gas production is essential for optimizing exploration strategies and guiding subsequent fracturing. However, in the new development of shale gas blocks, two main challenges are encountered: (1) data is insufficient, and (2) the dynamic production characteristics of shale gas wells, influenced by factors such as reservoirs and engineering, exhibit complex non-linear and non-stationary features. The inherent black-box nature of deep learning models raises concerns among decision-makers about the reliability of results. The current artificial intelligence model overlooks these factors, resulting in limitations in the accuracy and interpretability of the model. To address these problems, a novel domain adaptation methodology is proposed using physical constraints. First, production data from the source domain is segmented into multiple subdomains to enhance sample diversity. Subsequently, the positive transfer learning subdomains are identified by comparing maximum mean discrepancy (MMD) and global average distance metrics. Then, we integrate all transferable knowledge to create a more comprehensive target model. Finally, by incorporating drilling, completion, and geological data as physical constraints, we develop a hybrid model consisting of a multi-layer perceptron (MLP) and a Transformer, aiming to maximize interpretability, which is proved through comparison of symbolic transfer entropy (STE). The performance of the proposed method is experimentally validated on shale gas production data from two blocks in China. The average RMSE, MAE, and R2 on the target domain are 0.2454 (104 m3/d), 0.1552 (104 m3/d), and 0.88, respectively. These values are significantly superior to the traditional methods. Additionally, we demonstrate the superiority of our method in terms of causality through a comparison with Granger causality. The interpretability of static and dynamic data in the prediction process was studied using zero-value masking and attention mechanisms, respectively. Experimental results demonstrate the effectiveness and superiority of the proposed method for forecasting shale gas production under data insufficiency.

Altered Cortical Information Interaction During Respiratory Events in Children with Obstructive Sleep Apnea-Hypopnea Syndrome.

Obstructive sleep apnea-hypopnea syndrome (OSAHS) significantly impairs children's growth and cognition. This study aims to elucidate the pathophysiological mechanisms underlying OSAHS in children, with a particular focus on the alterations in cortical information interaction during respiratory events. We analyzed sleep electroencephalography before, during, and after events, utilizing Symbolic Transfer Entropy (STE) for brain network construction and information flow assessment. The results showed a significant increase in STE after events in specific frequency bands during N2 and rapid eye movement (REM) stages, along with increased STE during N3 stage events. Moreover, a noteworthy rise in the information flow imbalance within and between hemispheres was found after events, displaying unique patterns in central sleep apnea and hypopnea. Importantly, some of these alterations were correlated with symptom severity. These findings highlight significant changes in brain region coordination and communication during respiratory events, offering novel insights into OSAHS pathophysiology in children.

An efficient machine learning approach for extracting eSports players’ distinguishing features and classifying their skill levels using symbolic transfer entropy and consensus nested cross-validation

An efficient machine learning approach for extracting eSports players’ distinguishing features and classifying their skill levels using symbolic transfer entropy and consensus nested cross-validation

Electroencephalographic Features of Elderly Patients during Anesthesia Induction with Remimazolam: A Substudy of a Randomized Controlled Trial.

Although remimazolam is used as a general anesthetic in elderly patients due to its hemodynamic stability, the electroencephalogram characteristics of remimazolam are not well known. The purpose of this study was to identify the electroencephalographic features of remimazolam-induced unconsciousness in elderly patients and compare them with propofol. Remimazolam (n = 26) or propofol (n = 26) were randomly administered for anesthesia induction in surgical patients. The hypnotic agent was blinded only to the patients. During the induction of anesthesia, remimazolam was administered at a rate of 6 mg · kg-1 · h-1, and propofol was administered at a target effect-site concentration of 3.5 μg/ml. The electroencephalogram signals from eight channels (Fp1, Fp2, Fz, F3, F4, Pz, P3, and P4, referenced to A2, using the 10 to 20 system) were acquired during the induction of anesthesia and in the postoperative care unit. Power spectrum analysis was performed, and directed functional connectivity between frontal and parietal regions was evaluated using normalized symbolic transfer entropy. Functional connectivity in unconscious processes induced by remimazolam or propofol was compared with baseline. To compare each power of frequency over time of the two hypnotic agents, a permutation test with t statistic was conducted. Compared to the baseline in the alpha band, the feedback connectivity decreased by averages of 46% and 43%, respectively, after the loss of consciousness induced by remimazolam and propofol (95% CI for the mean difference: -0.073 to -0.044 for remimazolam [P < 0.001] and -0.068 to -0.042 for propofol [P < 0.001]). Asymmetry in the feedback and feedforward connectivity in the alpha band was suppressed after the loss of consciousness induced by remimazolam and propofol. There were no significant differences in the power of each frequency over time between the two hypnotic agents (minimum q value = 0.4235). Both regimens showed a greater decrease in feedback connectivity compared to a decrease in feedforward connectivity after loss of consciousness, leading to a disruption of asymmetry between the frontoparietal connectivity.

Root cause analysis of faults in cement pre-decomposition system using kernel principal component analysis and multi-scale symbolic transfer entropy

The industrial production process is complex, and when a fault occurs, multiple variables deviate from their normal state. In this paper, a solution is proposed which utilizes correlation and causality analysis to identify the root cause of failure at an early stage. Kernel principal component analysis (KPCA) demonstrates excellent monitoring performance in nonlinear systems. By applying the unified contribution graph method to KPCA, the main fault variables can be identified. Additionally, the Multi-Scale Symbolic Transfer Entropy (MSTE) method is employed to construct a delay causality diagram, addressing non-stationarity and lack of time-lag analysis. It considers time delays between variables while excluding indirect causality to diagnose the root failure point. The effectiveness of KPCA-MSTE is further validated by analyzing the failure process of a cement predecomposition system in actual production.

Predictions of the Key Operating Parameters in Waste Incineration Using Big Data and a Multiverse Optimizer Deep Learning Model

In order to accurately predict the key operating parameters of waste incinerators, this paper proposes a prediction method based on big data and a Multi-Verse Optimizer deep learning model, thus providing a powerful reference for controlling the optimization of the incinerator combustion process. The key operating parameters that were predicted, according to the control objectives, were determined to be the steam flow, gas oxygen, and flue temperature. Firstly, a large amount of measurement data were collected, and 27 relevant control system parameters with a high correlation with the predicted variables were obtained via a mechanism analysis. The input variables of the prediction model were further determined using the improved WesselN symbolic transfer entropy algorithm. The delay time between the variables was found using a gray correlation coefficient, the prediction time was determined to be 6 min according to the delay time distribution of the flame feature, and the time delay compensation was applied to each parameter. Finally, the support vector machine was optimized using a Multi-Verse Optimization algorithm to complete the prediction of the key operating parameters. Experiments showed that the root mean square error of the proposed model for the three output variables—the steam flow, gas oxygen, and flue temperature—were 0.3035, 0.2477, and 1.6773, respectively, which provides a high accuracy compared to other models.

Abnormal interaction between cortical regions of obstructive sleep apnea hypopnea syndrome children.

Obstructive sleep apnea hypopnea syndrome negatively affects the cognitive function of children. This study aims to find potential biomarkers for obstructive sleep apnea hypopnea syndrome in children by investigating the patterns of sleep electroencephalography networks. The participants included 16 mild obstructive sleep apnea hypopnea syndrome children, 12 severe obstructive sleep apnea hypopnea syndrome children, and 13 healthy controls. Effective brain networks were constructed using symbolic transfer entropy to assess cortical information interaction. The information flow pattern in the participants was evaluated using the parameters cross-within variation and the ratio of posterior-anterior information flow. Obstructive sleep apnea hypopnea syndrome children had a considerably higher symbolic transfer entropy in the full frequency band of N1, N2, and rapid eye movement (REM) stages (P < 0.05), and a significantly lower symbolic transfer entropy in full frequency band of N3 stage (P < 0.005), in comparison with the healthy controls. In addition, the cross-within variation of the β frequency band across all sleep stages were significantly lower in the obstructive sleep apnea hypopnea syndrome group than in the healthy controls (P < 0.05). What is more, the posterior-anterior information flowin the β frequency band of REM stage was significantly higher in mild obstructive sleep apnea hypopnea syndrome children than in the healthy controls (P < 0.05). These findings may serve as potential biomarkers for obstructive sleep apnea hypopnea syndrome in children and provide new insights into the pathophysiological mechanisms.

Nervous-System-Wise Functional Estimation of Directed Brain-Heart Interplay Through Microstate Occurrences.

The quantification of functional brain-heart interplay (BHI) through analysis of the dynamics of the central and autonomic nervous systems provides effective biomarkers for cognitive, emotional, and autonomic state changes. Several computational models have been proposed to estimate BHI, focusing on a single sensor, brain region, or frequency activity. However, no models currently provide a directional estimation of such interplay at the organ level. This study proposes an analysis framework to estimate BHI that quantifies the directional information flow between whole-brain and heartbeat dynamics. System-wise directed functional estimation is performed through an ad-hoc symbolic transfer entropy implementation, which leverages on EEG-derived microstate series and on partition of heart rate variability series. The proposed framework is validated on two different experimental datasets: the first investigates the cognitive workload performed through mental arithmetic and the second focuses on an autonomic maneuver using a cold pressor test (CPT). The experimental results highlight a significant bidirectional increase in BHI during cognitive workload with respect to the preceding resting phase and a higher descending interplay during a CPT compared to the preceding rest and following recovery phases. These changes are not detected by the intrinsic self entropy of isolated cortical and heartbeat dynamics. This study corroborates the literature on the BHI phenomenon under these experimental conditions and the new perspective provides novel insights from an organ-level viewpoint. A system-wise perspective of the BHI phenomenon may provide new insights into physiological and pathological processes that may not be completely understood at a lower level/scale of analysis.

Traceability of abnormal energy consumption modes in grinding systems based on evolution analysis of causal network structure

Abnormal energy consumption mode tracing is used to locate the root cause of the system deviating from the normal energy consumption mode (ECM), in order to support operators in understanding system abnormal changes and abnormal propagation trends. Due to the compensatory effect of closed-loop control systems on external interference and the influence of processes such as coarse powder regrinding and circulating air utilization during grinding, there is a strong correlation between parameters, and anomalies are easily propagated between parameters. Tracing the root cause variable of anomalies is challenging. Previous abnormal ECM tracing ignored the dynamic changes in causal relationships between coupled variables in the circulation process, and had problems with strong expert knowledge dependence and slow response. Therefore, this paper proposes an abnormal ECM tracing method based on causal network structure evolution analysis. It proposes directed symbolic transfer entropy to measure variable causal relationships, and combines the advantages of complex networks in characterizing complex system dynamic characteristics to construct an ECM causal network. By analyzing network structure characteristics, it locates the root cause variables of abnormal ECMs and infers propagation paths. Finally, through the application of six types of abnormal scenarios in actual slag grinding processes, it verifies the accuracy of the proposed method in locating abnormal roots and identifying propagation paths.

Kendall transfer entropy: a novel measure for estimating information transfer in complex systems

Objective. Transfer entropy (TE) has been widely used to infer causal relationships among dynamical systems, especially in neuroscience. Kendall transformation provides a novel quantization method for estimating information-theoretic measures and shows potential advantages for small-sample neural signals. But it has yet to be introduced into the framework of TE estimation, which commonly suffers from the limitation of small sample sizes. This paper aims to introduce the idea of Kendall correlation into TE estimation and verify its effect. Approach. We proposed the Kendall TE (KTE) which combines the improved Kendall transformation and the TE estimation. To confirm its effectiveness, we compared KTE with two common TE estimation techniques: the adaptive partitioning algorithm (D-V partitioning) and the symbolic TE. Their performances were estimated by simulation experiments which included linear, nonlinear, linear + nonlinear models and neural mass models. Moreover, the KTE was also applied to real electroencephalography (EEG) recordings to quantify the directional connectivity between frontal and parietal regions with propofol-induced general anesthesia. Main results. The simulation results showed that the KTE outperformed the other two methods by many measures: (1) identifying the coupling direction under a small sample size; (2) the sensitivity to coupling strength; (3) noise resistance; and (4) the sensitivity to time-dependent coupling changes. For real EEG recordings, the KTE clearly detected the disrupted frontal-to-parietal connectivity in propofol-induced unconsciousness, which is in agreement with previous findings. Significance. We reveal that the proposed KTE method is a robust and powerful tool for estimating TE, and is particularly suitable for small sample sizes. The KTE also provides an innovative form of quantizing continuous time series for information-theoretic measures.

Separation of responsive and unresponsive patients under clinical conditions: comparison of symbolic transfer entropy and permutation entropy

Electroencephalogram (EEG)-based monitoring during general anesthesia may help prevent harmful effects of high or low doses of general anesthetics. There is currently no convincing evidence in this regard for the proprietary algorithms of commercially available monitors. The purpose of this study was to investigate whether a more mechanism-based parameter of EEG analysis (symbolic transfer entropy, STE) can separate responsive from unresponsive patients better than a strictly probabilistic parameter (permutation entropy, PE) under clinical conditions. In this prospective single-center study, the EEG of 60 surgical ASA I–III patients was recorded perioperatively. During induction of and emergence from anesthesia, patients were asked to squeeze the investigators’ hand every 15s. Time of loss of responsiveness (LoR) during induction and return of responsiveness (RoR) during emergence from anesthesia were registered. PE and STE were calculated at −15s and +30s of LoR and RoR and their ability to separate responsive from unresponsive patients was evaluated using accuracy statistics. 56 patients were included in the final analysis. STE and PE values decreased during anesthesia induction and increased during emergence. Intra-individual consistency was higher during induction than during emergence. Accuracy values during LoR and RoR were 0.71 (0.62–0.79) and 0.60 (0.51–0.69), respectively for STE and 0.74 (0.66–0.82) and 0.62 (0.53–0.71), respectively for PE. For the combination of LoR and RoR, values were 0.65 (0.59–0.71) for STE and 0.68 (0.62–0.74) for PE. The ability to differentiate between the clinical status of (un)responsiveness did not significantly differ between STE and PE at any time. Mechanism-based EEG analysis did not improve differentiation of responsive from unresponsive patients compared to the probabilistic PE.Trial registration: German Clinical Trials Register ID: DRKS00030562, November 4, 2022, retrospectively registered.

Study on characteristic of epileptic multi-electroencephalograph base on Hilbert-Huang transform and brain network dynamics.

Lots of studies have been carried out on characteristic of epileptic Electroencephalograph (EEG). However, traditional EEG characteristic research methods lack exploration of spatial information. To study the characteristics of epileptic EEG signals from the perspective of the whole brain，this paper proposed combination methods of multi-channel characteristics from time-frequency and spatial domains. This paper was from two aspects: Firstly, signals were converted into 2D Hilbert Spectrum (HS) images which reflected the time-frequency characteristics by Hilbert-Huang Transform (HHT). These images were identified by Convolutional Neural Network (CNN) model whose sensitivity was 99.8%, accuracy was 98.7%, specificity was 97.4%, F1-score was 98.7%, and AUC-ROC was 99.9%. Secondly, the multi-channel signals were converted into brain networks which reflected the spatial characteristics by Symbolic Transfer Entropy (STE) among different channels EEG. And the results show that there are different network properties between ictal and interictal phase and the signals during the ictal enter the synchronization state more quickly, which was verified by Kuramoto model. To summarize, our results show that there was different characteristics among channels for the ictal and interictal phase, which can provide effective physical non-invasive indicators for the identification and prediction of epileptic seizures.

Multiscale Partial Symbolic Transfer Entropy for Time-Delay Root Cause Diagnosis in Nonstationary Industrial Processes

Nonstationary characteristics and lack of time delays analysis remain two large obstacles to diagnosing fault root cause in industrial processes. However, ignoring these two issues, many traditional methods tend to introduce some spurious or indirect causal relationships that interfere with root cause judgments. In this article, multiscale partial symbolic transfer entropy (MPSTE) is proposed to handle the above-mentioned problems. On the one hand, a multivariate multiscale filter module of MPSTE is designed to estimate the information transfer on multiple time scales and detect time delays while considering the indirect causality of the multivariate industrial system. On the other hand, to capture the causal information from the nonstationary processes, MPSTE introduces a segmental symbolization procedure and maps time series into finite rank vectors, overcoming the limitation of stationarity. Moreover, two causal criteria for MPSTE are constructed to identify the direction and significance level of causality correspondingly. MPSTE can thus provide sufficient information for causal relationships and exclude spurious causality. Finally, a causal diagram construction strategy with time delay based on MPSTE is established to diagnose the root cause and trace the fault propagation path. The results of experimental verification in a coal mill rig prove the effectiveness of the proposed strategy.

Discriminating between bipolar and major depressive disorder using a machine learning approach and resting-state EEG data

ObjectiveDistinguishing major depressive disorder (MDD) from bipolar disorder (BD) is a crucial clinical challenge as effective treatment is quite different for each condition. In this study electroencephalography (EEG) was explored as an objective biomarker for distinguishing MDD from BD using an efficient machine learning algorithm (MLA) trained by a relatively large and balanced dataset. MethodsA 3 step MLA was applied: (1) a multi-step preprocessing method was used to improve the quality of the EEG signal, (2) symbolic transfer entropy (STE), an effective connectivity measure, was applied to the resultant EEG and (3) the MLA used the extracted STE features to distinguish MDD (N = 71) from BD (N = 71) subjects. Results14 connectivity features were selected by the proposed algorithm. Most of the selected features were related to the frontal, parietal, and temporal lobe electrodes. The major involved regions were the Broca region in the frontal lobe and the somatosensory association cortex in the parietal lobe. These regions are near electrodes FC5 and CPz and are involved in processing language and sensory information, respectively. The resulting classifier delivered an evaluation accuracy of 88.5% and a test accuracy of 89.3%, using 80% of the data for training and evaluation and the remaining 20% for testing, respectively. ConclusionsThe high evaluation and test accuracies of our algorithm, derived from a large balanced training sample suggests that this method may hold significant promise as a clinical tool. SignificanceThe proposed MLA may provide an inexpensive and readily available tool that clinicians may use to enhance diagnostic accuracy and shorten time to effective treatment.

Universal Lifespan Trajectories of Source-Space Information Flow Extracted from Resting-State MEG Data.

Source activity was extracted from resting-state magnetoencephalography data of 103 subjects aged 18–60 years. The directionality of information flow was computed from the regional time courses using delay symbolic transfer entropy and phase entropy. The analysis yielded a dynamic source connectivity profile, disentangling the direction, strength, and time delay of the underlying causal interactions, producing independent time delays for cross-frequency amplitude-to-amplitude and phase-to-phase coupling. The computation of the dominant intrinsic coupling mode (DoCM) allowed me to estimate the probability distribution of the DoCM independently of phase and amplitude. The results support earlier observations of a posterior-to-anterior information flow for phase dynamics in {α1, α2, β, γ} and an opposite flow (anterior to posterior) in θ. Amplitude dynamics reveal posterior-to-anterior information flow in {α1, α2, γ}, a sensory-motor β-oriented pattern, and an anterior-to-posterior pattern in {δ, θ}. The DoCM between intra- and cross-frequency couplings (CFC) are reported here for the first time and independently for amplitude and phase; in both domains {δ, θ, α1}, frequencies are the main contributors to DoCM. Finally, a novel brain age index (BAI) is introduced, defined as the ratio of the probability distribution of inter- over intra-frequency couplings. This ratio shows a universal age trajectory: a rapid rise from the end of adolescence, reaching a peak in adulthood, and declining slowly thereafter. The universal pattern is seen in the BAI of each frequency studied and for both amplitude and phase domains. No such universal age dependence was previously reported.

Electroencephalogram and surface electromyogram fusion-based precise detection of lower limb voluntary movement using convolution neural network-long short-term memory model.

The electroencephalogram (EEG) and surface electromyogram (sEMG) fusion has been widely used in the detection of human movement intention for human–robot interaction, but the internal relationship of EEG and sEMG signals is not clear, so their fusion still has some shortcomings. A precise fusion method of EEG and sEMG using the CNN-LSTM model was investigated to detect lower limb voluntary movement in this study. At first, the EEG and sEMG signal processing of each stage was analyzed so that the response time difference between EEG and sEMG can be estimated to detect lower limb voluntary movement, and it can be calculated by the symbolic transfer entropy. Second, the data fusion and feature of EEG and sEMG were both used for obtaining a data matrix of the model, and a hybrid CNN-LSTM model was established for the EEG and sEMG-based decoding model of lower limb voluntary movement so that the estimated value of time difference was about 24 ∼ 26 ms, and the calculated value was between 25 and 45 ms. Finally, the offline experimental results showed that the accuracy of data fusion was significantly higher than feature fusion-based accuracy in 5-fold cross-validation, and the average accuracy of EEG and sEMG data fusion was more than 95%; the improved average accuracy for eliminating the response time difference between EEG and sEMG was about 0.7 ± 0.26% in data fusion. In the meantime, the online average accuracy of data fusion-based CNN-LSTM was more than 87% in all subjects. These results demonstrated that the time difference had an influence on the EEG and sEMG fusion to detect lower limb voluntary movement, and the proposed CNN-LSTM model can achieve high performance. This work provides a stable and reliable basis for human–robot interaction of the lower limb exoskeleton.

A Machine Learning Algorithm to Discriminating Between Bipolar and Major Depressive Disorders Based on Resting EEG Data.

Distinguishing major depressive disorder (MDD) from bipolar disorder (BD) is a crucial clinical challenge due to the lack of known biomarkers. Conventional methods of diagnosis rest exclusively on symptomatic presentation, and personal and family history. As a result, BD-depressed episode (BD-DE) is often misdiagnosed as MDD, and inappropriate therapy is given. Electroencephalography (EEG) has been widely studied as a potential source of biomarkers to differentiate these disorders. Previous attempts using machine learning (ML) methods have delivered insufficient sensitivity and specificity for clinical use, likely as a consequence of the small training set size, and inadequate ML methodology. We hope to overcome these limitations by employing a training dataset of resting-state EEG from 71 MDD and 71 BD patients. We introduce a robust 3 steps ML technique: 1) a multi-step preprocessing method is used to improve the quality of the EEG signal 2) symbolic transfer entropy (STE), which is an effective connectivity measure, is applied to the resultant EEG signals 3) the ML algorithm uses the extracted STE features to distinguish MDD from BD patients. Clinical Relevance--- The accuracy of our algorithm, derived from a large sample of patients, suggests that this method may hold significant promise as a clinical tool. The proposed method delivered total accuracy, sensitivity, and specificity of 84.9%, 83.4%, and 87.1%, respectively.

Minimal EEG channel selection for depression detection with connectivity features during sleep

Background and objectivesSleeping cortical electroencephalogram (EEG) has the potential for depression detection, for different sleep structure and cortical connection have been proved in depressed patients. However, the operation of multi-channel sleep EEG recording is cumbersome and requires laboratory equipment and professional sleep technician. Here, we focus on the depression detection using minimal sleep EEG channels. MethodsSixteen channels of EEG data of 30 patients with depression and 30 age-matched normal controls were recorded during sleep. Power spectral density of each single EEG channel was calculated, followed by measuring the symbolic transfer entropy (STE) and weighed phase lag index (WPLI) between EEG channel pairs in various frequency bands. Thereafter, these features were evaluated by F-score in the two-way classification (depression vs. control) of 30-s sleep EEG segments. Based on the F-score, entropy method was introduced to calculate the weight which could further assess the classification ability of various EEG channels or channel pairs. Finally, machine learning was implemented to verify the important EEG channels or channel pairs in depression diagnosis. ResultsThe features characterizing the inter-hemispheric connectivity in the posterior lobe, especially in the temporal lobe, showed high classification capacity. The classification accuracy of using two and four EEG channels in the temporal lobe were 97.96% and 99.61%, respectively. ConclusionsThis study showed the possibility of using only a few sleep EEG channels for depression screening, which may greatly facilitate the diagnosis of depression outside the hospital.