Detecting ventricular tachycardia and fibrillation by complexity measure

Prediction of the effect of electrical defibrillation by using spectral feature parameters

This paper proposes a system for predicting the effect of electrical defibrillation using spectral feature parameters. The proposed method consists of two-stage prediction. The first stage involves predicting whether electrical defibrillation is "Successful" or "Ineffective." As the next stage, if the proposed prediction system determines "Ineffective," the proposed system discriminates between "VF recurrence" or "Failure" for electrical defibrillation. To develop the prediction system, feature parameters for the target electrocardiograms (ECGs) were first extracted by using the wavelet transform and spectral analysis. Next, effective feature parameters for prediction are selected through an analysis of variance. Moreover, in the preprocessing phase, the Synthetic Minority Oversampling Technique method and standardization are introduced. Finally, support vector machines with some kernel functions and the regularization method are utilized to predict the three states, i.e., "Successful," "Failure," and "VF recurrence," for electrical defibrillation in two phases. In this paper, we present our analysis method for ECGs and evaluate the effectiveness of the proposed prediction system.

LDCNN: A new arrhythmia detection technique with ECG signals using a linear deep convolutional neural network

The electrocardiogram (ECG) is a fundamental and widely used tool for diagnosing cardiovascular diseases. It involves recording cardiac electrical signals using electrodes, which illustrate the functioning of cardiac muscles during contraction and relaxation phases. ECG is instrumental in identifying abnormal cardiac activity, heart attacks, and various cardiac conditions. Arrhythmia detection, a critical aspect of ECG analysis, entails accurately classifying heartbeats. However, ECG signal analysis demands a high level of expertise, introducing the possibility of human errors in interpretation. Hence, there is a clear need for robust automated detection techniques. Recently, numerous methods have emerged for arrhythmia detection from ECG signals. In our research, we developed a novel one-dimensional deep neural network technique called linear deep convolutional neural network (LDCNN) to identify arrhythmias from ECG signals. We compare our suggested method with several state-of-the-art algorithms for arrhythmia detection. We evaluate our methodology using benchmark datasets, including the PTB Diagnostic ECG and MIT-BIH Arrhythmia databases. Our proposed method achieves high accuracy rates of 99.24% on the PTB Diagnostic ECG dataset and 99.38% on the MIT-BIH Arrhythmia dataset.

Linear mixed-effect models for correlated response to process electroencephalogram recordings

A data set of clinical studies of electroencephalogram recordings (EEG) following data acquisition protocols in control individuals (Eyes Closed Wakefulness - Eyes Open Wakefulness, Hyperventilation, and Optostimulation) are quantified with information theory metrics, namely permutation Shanon entropy and permutation Lempel Ziv complexity, to identify functional changes. This work implement Linear mixed-effects models (LMEMs) for confirmatory hypothesis testing. The results show that EEGs have high variability for both metrics and there is a positive correlation between them. The mean of permutation Lempel-Ziv complexity and permutation Shanon entropy used simultaneously for each of the four states are distinguishable from each other. However, used separately, the differences between permutation Lempel-Ziv complexity or permutation Shanon entropy of some states were not statistically significant. This shows that the joint use of both metrics provides more information than the separate use of each of them. Despite their wide use in medicine, LMEMs have not been commonly applied to simultaneously model metrics that quantify EEG signals. Modeling EEGs using a model that characterizes more than one response variable and their possible correlations represents a new way of analyzing EEG data in neuroscience.

Classifier for the functional state of the respiratory system via descriptors determined by using multimodal technology

Currently, intelligent systems built on a multimodal basis are used to study the functional state of living objects. Its essence lies in the fact that a decision is made through several independent information channels with the subsequent aggregation of these decisions. The method of forming descriptors for classifiers of the functional state of the respiratory system includes the study of the spectral range of the respiratory rhythm and the construction of the wavelet plane of the monitoring electrocardiosignal overlapping this range. Then, variations in the breathing rhythm are determined along the corresponding lines of the wavelet plane. Its analysis makes it possible to select slow waves corresponding to the breathing rhythm and systemic waves of the second order. Analysis of the spectral characteristics of these waves makes it possible to form a space of informative features for classifiers of the functional state of the respiratory system. To construct classifiers of the functional state of the respiratory system, hierarchical classifiers were used. As an example, we took a group of patients with pneumonia with a well-defined diagnosis (radiography, X-ray tomography, laboratory data) and a group of volunteers without pulmonary pathology. The diagnostic sensitivity of the obtained classifier was 76% specificity with a diagnostic specificity of 82%, which is comparable to the results of X-ray studies. It is shown that the corresponding lines of the wavelet planes are correlated with the respiratory system and, using their Fourier analysis, descriptors can be obtained for training neural network classifiers of the functional state of the respiratory system.

A VLSI Chip for the Abnormal Heart Beat Detection Using Convolutional Neural Network

The heart is one of the human body’s vital organs. An electrocardiogram (ECG) provides continuous tracings of the electrophysiological activity originated from heart, thus being widely used for a variety of diagnostic purposes. This study aims to design and realize an artificial intelligence (AI)-based abnormal heart beat detection with applications for early detection and timely treatment for heart diseases. A convolutional neural network (CNN) was employed to achieve a fast and accurate identification. In order to meet the requirements of the modularity and scalability of the circuit, modular and efficient processing element (PE) units and activation function modules were designed. The proposed CNN was implemented using a TSMC 0.18 μm CMOS technology and had an operating frequency of 60 MHz with chip area of 1.42 mm2 and maximum power dissipation of 4.4 mW. Furthermore, six types of ECG signals drawn from the MIT-BIH arrhythmia database were used for performance evaluation. Results produced by the proposed hardware showed that the discrimination rate was 96.3% with high efficiency in calculation, suggesting that it may be suitable for wearable devices in healthcare.

Non-linear Methods Predominant in Fetal Heart Rate Analysis: A Systematic Review

The analysis of fetal heart rate variability has served as a scientific and diagnostic tool to quantify cardiac activity fluctuations, being good indicators of fetal well-being. Many mathematical analyses were proposed to evaluate fetal heart rate variability. We focused on non-linear analysis based on concepts of chaos, fractality, and complexity: entropies, compression, fractal analysis, and wavelets. These methods have been successfully applied in the signal processing phase and increase knowledge about cardiovascular dynamics in healthy and pathological fetuses. This review summarizes those methods and investigates how non-linear measures are related to each paper's research objectives. Of the 388 articles obtained in the PubMed/Medline database and of the 421 articles in the Web of Science database, 270 articles were included in the review after all exclusion criteria were applied. While approximate entropy is the most used method in classification papers, in signal processing, the most used non-linear method was Daubechies wavelets. The top five primary research objectives covered by the selected papers were detection of signal processing, hypoxia, maturation or gestational age, intrauterine growth restriction, and fetal distress. This review shows that non-linear indices can be used to assess numerous prenatal conditions. However, they are not yet applied in clinical practice due to some critical concerns. Some studies show that the combination of several linear and non-linear indices would be ideal for improving the analysis of the fetus's well-being. Future studies should narrow the research question so a meta-analysis could be performed, probing the indices' performance.

Complex Relationship Between Daily Partner Violence and Alcohol Use Among Violent Heterosexual Men

Although alcohol use and partner violence are consistently associated, the nature of the alcohol-violence relationship is still unclear. The purpose of this pilot study was to use longitudinal daily assessments of male partners' alcohol use and violent events to identify the nature of the alcohol-violence relationship, employing both linear and nonlinear analyses. The participants were 20 adult heterosexual couples of whom the woman reported experiencing partner violence in the prior 30 days. Each partner provided a separate daily telephone report for 8 weeks via an automated interactive voice response (IVR), concerning the previous day's violence, alcohol use, stressors, emotional reactions, and concerns for children. Individual IVR databases were merged to form a combined couple's IVR time series. Time series were analyzed using graphic, linear, and nonlinear methods. Graphic analysis using state space grids found no consistent pattern across couples. Similarly, linear analysis using same-day cross-correlation and prior-day beta statistics found no significant group-level alcohol-violence relationship. Using cross-approximate entropy statistics and differential structural equation modeling, no nonlinear relationships between alcohol use and violence were noted either. Whether applying linear or nonlinear analytic methods, there is no group-level relationship between alcohol use by male perpetrators and their violent acts. The implications are significant. First, the alcohol-violence relationship may differ among subgroups. Second, couples need to be assessed thoroughly to determine their unique relationship with alcohol use, so that couple-specific interventions can be designed. Third, if perpetrators believe that their violence is facilitated by their alcohol use, then alcohol reduction should be encouraged despite any evidence suggesting a different alcohol-violence relationship. Finally, the accepted alcohol-causes-violence belief held by many providers needs to be reconsidered. Because the nature of the alcohol-violence relationship varies considerably across couples, clinicians should seek to understand their unique relationship applying across-the-board management approaches.

Reconstructed State Space Features for Classification of ECG Signals

Background:

Cardiac arrhythmias are considered as one of the most serious health conditions; therefore, accurate and quick diagnosis of these conditions is highly paramount for the electrocardiogram (ECG) signals. Moreover, are rather difficult for the cardiologists to diagnose with unaided eyes due to a close similarity of these signals in the time domain.

Objective:

In this paper, an image-based and machine learning method were presented in order to investigate the differences between the three cardiac arrhythmias of VF, VT, SVT and the normal signal.

Material and Methods:

In this simulation study, the ECG data used are collected from 3 databases, including Boston Beth University Arrhythmias Center, Creighton University, and MIT-BIH. The proposed algorithm was implemented using MATLAB R2015a software and its simulation. At first, the signal is transmitted to the state space using an optimal time delay. Then, the optimal delay values are obtained using the particle swarm optimization algorithm and normalized mutual information criterion. Furthermore, the result is considered as a binary image. Then, 19 features are extracted from the image and the results are presented in the multilayer perceptron neural network for the purpose of training and testing.

Results:

In order to classify N-VF, VT-SVT, N-SVT, VF-VT, VT-N-VF, N-SVT-VF, VT-VF-SVT and VT-VF-SVT-N in the conducted experiments, the accuracy rates were determined at 99.5%, 100%, 94.98%, 100%,100%, 100%, 99.5%, 96.5% and 95%, respectively.

Conclusion:

In this paper, a new approach was developed to classify the abnormal signals obtained from an ECG such as VT, VF, and SVT compared to a normal signal. Compared to Other related studies, our proposed system significantly performed better

Is Perceived Need for Action Among Women in Violent Relationships Nonlinear and, If So, Why?

Despite the prevalence and impact of partner violence, we understand little about women's action taking except that it seems an unpredictable, nonlinear process. This article determines the degree of nonlinearity in perceived need for help, legal action, or leaving among women in violent relationships. The participants included 143 women who experienced violence in the previous month, enrolled from six primary care clinics. Baseline surveys assessed background characteristics and factors which may affect perceived need for action. Multiple times series assessments of violence and need for action were collected daily for 8 weeks via telephone Interactive Voice Response. Measures of nonlinearity of violence, perceived need for help, legal action, and leaving were computed. Repeated measures ANOVA assessed differences across measures of nonlinearity. To identify factors contributing to nonlinearity, staged multiple regression assessed the relationship between nonlinearity measures and outcomes. Ninety-three women completed sufficient time series for nonlinearity assessment. Measures of nonlinearity were lower for need for legal action compared with needs for help and leaving. Regression analysis suggested that isolation, social networks, and lack of awareness contribute to nonlinearity. Women's perceived need for legal action and its level of nonlinearity were lowest compared with those of help seeking and leaving. Although its relative linearity suggests that the need for legal action may be the most predictable, its lower mean rating suggests that legal action is a low priority. Although need for help and leaving are of higher priorities, their nonlinearity suggests that intervention will not yield predictable results.

Fully Convolutional Deep Neural Networks with Optimized Hyperparameters for Detection of Shockable and Non-Shockable Rhythms

Deep neural networks (DNN) are state-of-the-art machine learning algorithms that can be learned to self-extract significant features of the electrocardiogram (ECG) and can generally provide high-output diagnostic accuracy if subjected to robust training and optimization on large datasets at high computational cost. So far, limited research and optimization of DNNs in shock advisory systems is found on large ECG arrhythmia databases from out-of-hospital cardiac arrests (OHCA). The objective of this study is to optimize the hyperparameters (HPs) of deep convolutional neural networks (CNN) for detection of shockable (Sh) and nonshockable (NSh) rhythms, and to validate the best HP settings for short and long analysis durations (2-10 s). Large numbers of (Sh + NSh) ECG samples were used for training (720 + 3170) and validation (739 + 5921) from Holters and defibrillators in OHCA. An end-to-end deep CNN architecture was implemented with one-lead raw ECG input layer (5 s, 125 Hz, 2.5 uV/LSB), configurable number of 5 to 23 hidden layers and output layer with diagnostic probability p ∈ [0: Sh,1: NSh]. The hidden layers contain N convolutional blocks × 3 layers (Conv1D (filters = Fi, kernel size = Ki), max-pooling (pool size = 2), dropout (rate = 0.3)), one global max-pooling and one dense layer. Random search optimization of HPs = {N, Fi, Ki}, i = 1, … N in a large grid of N = [1, 2, … 7], Fi = [5;50], Ki = [5;100] was performed. During training, the model with maximal balanced accuracy BAC = (Sensitivity + Specificity)/2 over 400 epochs was stored. The optimization principle is based on finding the common HPs space of a few top-ranked models and prediction of a robust HP setting by their median value. The optimal models for 1-7 CNN layers were trained with different learning rates LR = [10-5; 10-2] and the best model was finally validated on 2-10 s analysis durations. A number of 4216 random search models were trained. The optimal models with more than three convolutional layers did not exhibit substantial differences in performance BAC = (99.31-99.5%). Among them, the best model was found with {N = 5, Fi = {20, 15, 15, 10, 5}, Ki = {10, 10, 10, 10, 10}, 7521 trainable parameters} with maximal validation performance for 5-s analysis (BAC = 99.5%, Se = 99.6%, Sp = 99.4%) and tolerable drop in performance (<2% points) for very short 2-s analysis (BAC = 98.2%, Se = 97.6%, Sp = 98.7%). DNN application in future-generation shock advisory systems can improve the detection performance of Sh and NSh rhythms and can considerably shorten the analysis duration complying with resuscitation guidelines for minimal hands-off pauses.

ECG-based Random Forest Classifier for Cardiac Arrest Rhythms

Rhythm annotation of out-of-hospital cardiac episodes (OHCA) is key for a better understanding of the interplay between resuscitation therapy and OHCA patient outcome. OHCA rhythms are classified in five categories, asystole (AS), pulseless electrical activity (PEA), pulsed rhythms (PR), ventricular fibrillation (VF) and ventricular tachycardia (VT). Manual OHCA annotation by expert clinicians is onerous and time consuming, so there is a need for accurate and automatic OHCA rhythm annotation methods. For this study 852 OHCA episodes of patients treated with Automated External Defibrillators (AED) by the Emergency Medical Services of the Basque Country were analyzed. Six expert clinicians reviewed the electrocardiogram (ECG) of 4214 AED rhythm analyses and annotated the rhythm. Their consensus decision was used as ground truth. There were a total of 2418 AS, 294 PR, 1008 PEA, 472 VF and 22 VT. The ECG analysis intervals were extracted and used to develop an automatic rhythm annotator. Data was partitioned patient-wise into training (70%) and test (30%). Performance was evaluated in terms of per class sensitivity (Se) and F-score (F1). The unweighted mean of sensitivity (UMS) and F-score were used as global performance metrics. The classification method is composed of a feature extraction and denoising stage based on the stationary wavelet transform of the ECG, and on a random forest classifier. The best model presented a per rhythm Se/F1 of 95.8/95.7, 43.3/52.2, 85.3/81.3, 94.2/96.1, 81.9/72.2 for AS, PR, PEA, VF and VT, respectively. The UMS for the test set was 80.2%, 2-points above that of previous solutions. This method could be used to retrospectively annotate large OHCA datasets and ameliorate the workload of manual OHCA rhythm annotation.

Complexity Changes in the US and China's Stock Markets: Differences, Causes, and Wider Social Implications

How different are the emerging and the well-developed stock markets in terms of efficiency? To gain insights into this question, we compared an important emerging market, the Chinese stock market, and the largest and the most developed market, the US stock market. Specifically, we computed the Lempel–Ziv complexity (LZ) and the permutation entropy (PE) from two composite stock indices, the Shanghai stock exchange composite index (SSE) and the Dow Jones industrial average (DJIA), for both low-frequency (daily) and high-frequency (minute-to-minute)stock index data. We found that the US market is basically fully random and consistent with efficient market hypothesis (EMH), irrespective of whether low- or high-frequency stock index data are used. The Chinese market is also largely consistent with the EMH when low-frequency data are used. However, a completely different picture emerges when the high-frequency stock index data are used, irrespective of whether the LZ or PE is computed. In particular, the PE decreases substantially in two significant time windows, each encompassing a rapid market rise and then a few gigantic stock crashes. To gain further insights into the causes of the difference in the complexity changes in the two markets, we computed the Hurst parameter H from the high-frequency stock index data of the two markets and examined their temporal variations. We found that in stark contrast with the US market, whose H is always close to 1/2, which indicates fully random behavior, for the Chinese market, H deviates from 1/2 significantly for time scales up to about 10 min within a day, and varies systemically similar to the PE for time scales from about 10 min to a day. This opens the door for large-scale collective behavior to occur in the Chinese market, including herding behavior and large-scale manipulation as a result of inside information.

Schizophrenia diagnosis using innovative EEG feature-level fusion schemes

Electroencephalogram (EEG) has become a practical tool for monitoring and diagnosing pathological/psychological brain states. To date, an increasing number of investigations considered differences between brain dynamic of patients with schizophrenia and healthy controls. However, insufficient studies have been performed to provide an intelligent and accurate system that detects the schizophrenia using EEG signals. This paper concerns this issue by providing new feature-level fusion algorithms. Firstly, we analyze EEG dynamics using three well-known nonlinear measures, including complexity (Cx), Higuchi fractal dimension (HFD), and Lyapunov exponents (Lya). Next, we propose some innovative feature-level fusion strategies to combine the information of these indices. We evaluate the effect of the classifier parameter (σ) adjustment and the cross-validation partitioning criteria on classification accuracy. The performance of EEG classification using combined features was compared with the non-combined attributes. Experimental results showed higher classification accuracy when feature-level features were utilized, compared to when each feature was used individually or all fed to the classifier simultaneously. Using the proposed algorithm, the classification accuracy increased up to 100%. These results establish the suggested framework as a superior scheme compared to the state-of-the-art EEG schizophrenia diagnosis tool.

Markov Models for Detection of Ventricular Arrhythmia

The advent of portable cardiac monitoring devices has enabled real-time analysis of cardiac signals. These devices can be used to develop algorithms for real-time detection of dangerous heart rhythms such as ventricular arrhythmias. This paper presents a Markov model based algorithm for real-time detection of ventricular tachycardia, ventricular flutter, and ventricular fibrillation episodes. The algorithm does not rely on any noise removal pre-processing or peak annotation of the original signal. When evaluated using ECG signals from three publicly available databases, the model resulted in an AUC of 0.96 and F1-score of 0.91 for 5-second long signals and an AUC of 0.97 and F1-score of 0.93 for 2-second long signals.

Mixed convolutional and long short-term memory network for the detection of lethal ventricular arrhythmia

Early defibrillation by an automated external defibrillator (AED) is key for the survival of out-of-hospital cardiac arrest (OHCA) patients. ECG feature extraction and machine learning have been successfully used to detect ventricular fibrillation (VF) in AED shock decision algorithms. Recently, deep learning architectures based on 1D Convolutional Neural Networks (CNN) have been proposed for this task. This study introduces a deep learning architecture based on 1D-CNN layers and a Long Short-Term Memory (LSTM) network for the detection of VF. Two datasets were used, one from public repositories of Holter recordings captured at the onset of the arrhythmia, and a second from OHCA patients obtained minutes after the onset of the arrest. Data was partitioned patient-wise into training (80%) to design the classifiers, and test (20%) to report the results. The proposed architecture was compared to 1D-CNN only deep learners, and to a classical approach based on VF-detection features and a support vector machine (SVM) classifier. The algorithms were evaluated in terms of balanced accuracy (BAC), the unweighted mean of the sensitivity (Se) and specificity (Sp). The BAC, Se, and Sp of the architecture for 4-s ECG segments was 99.3%, 99.7%, and 98.9% for the public data, and 98.0%, 99.2%, and 96.7% for OHCA data. The proposed architecture outperformed all other classifiers by at least 0.3-points in BAC in the public data, and by 2.2-points in the OHCA data. The architecture met the 95% Sp and 90% Se requirements of the American Heart Association in both datasets for segment lengths as short as 3-s. This is, to the best of our knowledge, the most accurate VF detection algorithm to date, especially on OHCA data, and it would enable an accurate shock no shock diagnosis in a very short time.

VENTRICULAR TACHYCARDIA AND FIBRILLATION DETECTION USING DWT AND DECISION TREE CLASSIFIER

Ventricular tachycardia (VT) and ventricular fibrillation (VF) are the life-threatening ventricular arrhythmias that require treatment in an emergency. Detection of VT and VF at an early stage is crucial for achieving the success of the defibrillation treatment. Hence an automatic system using computer-aided diagnosis tool is helpful in detecting the ventricular arrhythmias in electrocardiogram (ECG) signal. In this paper, a discrete wavelet transform (DWT) was used to denoise and decompose the ECG signals into different consecutive frequency bands to reduce noise. The methodology was tested using ECG data from standard CU ventricular tachyarrhythmia database (CUDB) and MIT-BIH malignant ventricular ectopy database (VFDB) datasets of PhysioNet databases. A set of time-frequency features consists of temporal, spectral, and statistical were extracted and ranked by the correlation attribute evaluation with ranker search method in order to improve the accuracy of detection. The ranked features were classified for VT and VF conditions using support vector machine (SVM) and decision tree (C4.5) classifier. The proposed DWT based features yielded the average sensitivity of 98%, specificity of 99.32%, and accuracy of 99.23% using a decision tree (C4.5) classifier. These results were better than the SVM classifier having an average accuracy of 92.43%. The obtained results prove that using DWT based time-frequency features with decision tree (C4.5) classifier can be one of the best choices for clinicians for precise detection of ventricular arrhythmias.

Multi-domain analysis of microvascular flow motion dynamics in NAFLD

OBJECTIVE

To determine whether analysis of microvascular network perfusion using complexity-based methods can discriminate between groups of individuals at an increased risk of developing CVD.

METHODS

Data were obtained from laser Doppler recordings of skin blood flux at the forearm in 50 participants with non-alcoholic fatty liver disease grouped for absence (n = 28) or presence (n = 14) of type 2 diabetes and use of calcium channel blocker medication (n = 8). Power spectral density was evaluated and Lempel-Ziv complexity determined to quantify signal information content at single and multiple time-scales to account for the different processes modulating network perfusion.

RESULTS

Complexity was associated with dilatory capacity and respiration and negatively with baseline blood flux and cardiac band power. The relationship between the modulators of flowmotion and complexity of blood flux is shown to change with time-scale improving discrimination between groups. Multiscale Lempel-Ziv achieved best classification accuracy of 86.1%.

CONCLUSIONS

Time and frequency domain measures alone are insufficient to discriminate between groups. As cardiovascular disease risk increases, the degree of complexity of the blood flux signal reduces, indicative of a reduced temporal activity and heterogeneous distribution of blood flow within the microvascular network sampled. Complexity-based methods, particularly multiscale variants, are shown to have good discriminatory capabilities.

A deep neural network learning algorithm outperforms a conventional algorithm for emergency department electrocardiogram interpretation

BACKGROUND

Cardiologs® has developed the first electrocardiogram (ECG) algorithm that uses a deep neural network (DNN) for full 12‑lead ECG analysis, including rhythm, QRS and ST-T-U waves. We compared the accuracy of the first version of Cardiologs® DNN algorithm to the Mortara/Veritas® conventional algorithm in emergency department (ED) ECGs.

METHODS

Individual ECG diagnoses were prospectively mapped to one of 16 pre-specified groups of ECG diagnoses, which were further classified as "major" ECG abnormality or not. Automated interpretations were compared to blinded experts'. The primary outcome was the performance of the algorithms in finding at least one "major" abnormality. The secondary outcome was the proportion of all ECGs for which all groups were identified, with no false negative or false positive groups ("accurate ECG interpretation"). Additionally, we measured sensitivity and positive predictive value (PPV) for any abnormal group.

RESULTS

Cardiologs® vs. Veritas® accuracy for finding a major abnormality was 92.2% vs. 87.2% (p < 0.0001), with comparable sensitivity (88.7% vs. 92.0%, p = 0.086), improved specificity (94.0% vs. 84.7%, p < 0.0001) and improved positive predictive value (PPV 88.2% vs. 75.4%, p < 0.0001). Cardiologs® had accurate ECG interpretation for 72.0% (95% CI: 69.6-74.2) of ECGs vs. 59.8% (57.3-62.3) for Veritas® (P < 0.0001). Sensitivity for any abnormal group for Cardiologs® and Veritas®, respectively, was 69.6% (95CI 66.7-72.3) vs. 68.3% (95CI 65.3-71.1) (NS). Positive Predictive Value was 74.0% (71.1-76.7) for Cardiologs® vs. 56.5% (53.7-59.3) for Veritas® (P < 0.0001).

CONCLUSION

Cardiologs' DNN was more accurate and specific in identifying ECGs with at least one major abnormal group. It had a significantly higher rate of accurate ECG interpretation, with similar sensitivity and higher PPV.

A Machine Learning Shock Decision Algorithm for Use During Piston-Driven Chest Compressions

GOAL

Accurate shock decision methods during piston-driven cardiopulmonary resuscitation (CPR) would contribute to improve therapy and increase cardiac arrest survival rates. The best current methods are computationally demanding, and their accuracy could be improved. The objective of this work was to introduce a computationally efficient algorithm for shock decision during piston-driven CPR with increased accuracy.

METHODS

The study dataset contains 201 shockable and 844 nonshockable ECG segments from 230 cardiac arrest patients treated with the LUCAS-2 mechanical CPR device. Compression artifacts were removed using the state-of-the-art adaptive filters, and shock/no-shock discrimination features were extracted from the stationary wavelet transform analysis of the filtered ECG, and fed to a support vector machine (SVM) classifier. Quasi-stratified patient wise nested cross-validation was used for feature selection and SVM hyperparameter optimization. The procedure was repeated 50 times to statistically characterize the results.

RESULTS

Best results were obtained for a six-feature classifier with mean (standard deviation) sensitivity, specificity, and total accuracy of 97.5 (0.4), 98.2 (0.4), and 98.1 (0.3), respectively. The algorithm presented a five-fold reduction in computational demands when compared to the best available methods, while improving their balanced accuracy by 3 points.

CONCLUSIONS

The accuracy of the best available methods was improved while drastically reducing the computational demands.

SIGNIFICANCE

An efficient and accurate method for shock decisions during mechanical CPR is now available to improve therapy and contribute to increase cardiac arrest survival.

EVALUATION OF SIGNAL PROCESSING TECHNIQUES IN DISCRIMINATING ECG SIGNALS OF MEN AND WOMEN DURING REST CONDITION AND EMOTIONAL STATES

Great range of electrocardiogram (ECG) signal processing methods can be found in the literature. In addition, the importance of gender differences in physiological activities was also identified in various conditions. This article aims to provide a comprehensive evaluation of linear and nonlinear ECG parameters to indicate suitable signal processing approaches which can show significant differences between men and women. These differences were investigated in two conditions: (i) during rest condition, and (ii) during the affective image inducements. A wide range of parameters from time-, frequency-, wavelet-, and nonlinear-techniques were examined. Applying the Wilcoxon rank sum test, significant differences between two genders were inspected. The analysis was performed on 47 college students at rest condition and while subjects watching four types of affective pictures, including sadness, happiness, fear, and peacefulness. The impact of these emotions on the results was also investigated. The results indicated that 72.95% and 72.61% of all features were significantly different between male and female in rest condition and affective inducements, respectively. In addition, the highest percentage of the significant difference between ECG parameters of men and women was achieved using nonlinear characteristics. Considering all features together, the highest significant difference between two genders was achieved for negative emotions, including sadness and fear. In conclusion, the results of this study emphasized the importance of gender role in cardiac responses during rest condition and different emotional states. Since these gender differences are well manifested by nonlinear signal processing techniques, dynamical gender-specific ECG system may improve the automatic emotion recognition accuracies.

Automated Method for Discrimination of Arrhythmias Using Time, Frequency, and Nonlinear Features of Electrocardiogram Signals

We developed an automated approach to differentiate between different types of arrhythmic episodes in electrocardiogram (ECG) signals, because, in real-life scenarios, a software application does not know in advance the type of arrhythmia a patient experiences. Our approach has four main stages: (1) Classification of ventricular fibrillation (VF) versus non-VF segments—including atrial fibrillation (AF), ventricular tachycardia (VT), normal sinus rhythm (NSR), and sinus arrhythmias, such as bigeminy, trigeminy, quadrigeminy, couplet, triplet—using four image-based phase plot features, one frequency domain feature, and the Shannon entropy index. (2) Classification of AF versus non-AF segments. (3) Premature ventricular contraction (PVC) detection on every non-AF segment, using a time domain feature, a frequency domain feature, and two features that characterize the nonlinearity of the data. (4) Determination of the PVC patterns, if present, to categorize distinct types of sinus arrhythmias and NSR. We used the Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) arrhythmia database, Creighton University’s VT arrhythmia database, the MIT-BIH atrial fibrillation database, and the MIT-BIH malignant ventricular arrhythmia database to test our algorithm. Binary decision tree (BDT) and support vector machine (SVM) classifiers were used in both stage 1 and stage 3. We also compared our proposed algorithm’s performance to other published algorithms. Our VF detection algorithm was accurate, as in balanced datasets (and unbalanced, in parentheses) it provided an accuracy of 95.1% (97.1%), sensitivity of 94.5% (91.1%), and specificity of 94.2% (98.2%). The AF detection was accurate, as the sensitivity and specificity in balanced datasets (and unbalanced, in parentheses) were found to be 97.8% (98.6%) and 97.21% (97.1%), respectively. Our PVC detection algorithm was also robust, as the accuracy, sensitivity, and specificity were found to be 99% (98.1%), 98.0% (96.2%), and 98.4% (99.4%), respectively, for balanced and (unbalanced) datasets.

Detection of Life Threatening Ventricular Arrhythmia Using Digital Taylor Fourier Transform

Accurate detection and classification of life-threatening ventricular arrhythmia episodes such as ventricular fibrillation (VF) and rapid ventricular tachycardia (VT) from electrocardiogram (ECG) is a challenging problem for patient monitoring and defibrillation therapy. This paper introduces a novel method for detection and classification of life-threatening ventricular arrhythmia episodes. The ECG signal is decomposed into various oscillatory modes using digital Taylor-Fourier transform (DTFT). The magnitude feature and a novel phase feature namely the phase difference (PD) are evaluated from the mode Taylor-Fourier coefficients of ECG signal. The least square support vector machine (LS-SVM) classifier with linear and radial basis function (RBF) kernels is employed for detection and classification of VT vs. VF, non-shock vs. shock and VF vs. non-VF arrhythmia episodes. The accuracy, sensitivity, and specificity values obtained using the proposed method are 89.81, 86.38, and 93.97%, respectively for the classification of Non-VF and VF episodes. Comparison with the performance of the state-of-the-art features demonstrate the advantages of the proposition.

Complexity Analysis of Global Temperature Time Series

Climate has complex dynamics due to the plethora of phenomena underlying its evolution. These characteristics pose challenges to conducting solid quantitative analysis and reaching assertive conclusions. In this paper, the global temperature time series (TTS) is viewed as a manifestation of the climate evolution, and its complexity is calculated by means of four different indices, namely the Lempel–Ziv complexity, sample entropy, signal harmonics power ratio, and fractal dimension. In the first phase, the monthly mean TTS is pre-processed by means of empirical mode decomposition, and the TTS trend is calculated. In the second phase, the complexity of the detrended signals is estimated. The four indices capture distinct features of the TTS dynamics in a 4-dim space. Hierarchical clustering is adopted for dimensional reduction and visualization in the 2-dim space. The results show that TTS complexity exhibits space-time variability, suggesting the presence of distinct climate forcing processes in both dimensions. Numerical examples with real-world data demonstrate the effectiveness of the approach.

Association between increased EEG signal complexity and cannabis dependence

Both acute and regular cannabis use affects the functioning of the brain. While several studies have demonstrated that regular cannabis use can impair the capacity to synchronize neural assemblies during specific tasks, less is known about spontaneous brain activity. This can be explored by measuring EEG complexity, which reflects the spontaneous variability of human brain activity. A recent study has shown that acute cannabis use can affect that complexity. Since the characteristics of cannabis use can affect the impact on brain functioning, this study sets out to measure EEG complexity in regular cannabis users with or without dependence, in comparison with healthy controls. We recruited 26 healthy controls, 25 cannabis users without cannabis dependence and 14 cannabis users with cannabis dependence, based on DSM IV TR criteria. The EEG signal was extracted from at least 250 epochs of the 500ms pre-stimulation phase during a visual evoked potential paradigm. Brain complexity was estimated using Lempel-Ziv Complexity (LZC), which was compared across groups by non-parametric Kruskall-Wallis ANOVA. The analysis revealed a significant difference between the groups, with higher LZC in participants with cannabis dependence than in non-dependent cannabis users. There was no specific localization of this effect across electrodes. We showed that cannabis dependence is associated to an increased spontaneous brain complexity in regular users. This result is in line with previous results in acute cannabis users. It may reflect increased randomness of neural activity in cannabis dependence. Future studies should explore whether this effect is permanent or diminishes with cannabis cessation.

SHOCKABLE VERSUS NONSHOCKABLE LIFE-THREATENING VENTRICULAR ARRHYTHMIAS USING DWT AND NONLINEAR FEATURES OF ECG SIGNALS

Shockable ventricular arrhythmias (VAs) such as ventricular tachycardia (VT) and ventricular fibrillation (VFib) are the life-threatening conditions requiring immediate attention. Cardiopulmonary resuscitation (CPR) and defibrillation are the significant immediate recommended treatments for these shockable arrhythmias to obtain the return of spontaneous circulation. However, accurate classification of these shockable VAs from nonshockable ones is the key step during defibrillation by automated external defibrillator (AED). Therefore, in this work, we have proposed a novel algorithm for an automated differentiation of shockable and nonshockable VAs from electrocardiogram (ECG) signal. The ECG signals are segmented into 5, 8 and 10[Formula: see text]s. These segmented ECGs are subjected to four levels of discrete wavelet transformation (DWT). Various nonlinear features such as approximate entropy ([Formula: see text], signal energy ([Formula: see text]), Fuzzy entropy ([Formula: see text]), Kolmogorov Sinai entropy ([Formula: see text], permutation entropy ([Formula: see text]), Renyi entropy ([Formula: see text]), sample entropy ([Formula: see text]), Shannon entropy ([Formula: see text]), Tsallis entropy ([Formula: see text]), wavelet entropy ([Formula: see text]), fractal dimension ([Formula: see text]), Kolmogorov complexity ([Formula: see text]), largest Lyapunov exponent ([Formula: see text]), recurrence quantification analysis (RQA) parameters ([Formula: see text]), Hurst exponent ([Formula: see text]), activity entropy ([Formula: see text]), Hjorth complexity ([Formula: see text]), Hjorth mobility ([Formula: see text]), modified multi scale entropy ([Formula: see text]) and higher order statistics (HOS) bispectrum ([Formula: see text]) are obtained from the DWT coefficients. Later, these features are subjected to sequential forward feature selection (SFS) method and selected features are then ranked using seven ranking methods namely, Bhattacharyya distance, entropy, Fuzzy maximum relevancy and minimum redundancy (mRMR), receiver operating characteristic (ROC), Student’s [Formula: see text]-test, Wilcoxon and ReliefF. These ranked features are supplied independently into the [Formula: see text]-Nearest Neighbor (kNN) classifier. Our proposed system achieved maximum accuracy, sensitivity and specificity of (i) 97.72%, 94.79% and 98.74% for 5[Formula: see text]s, (ii) 98.34%, 95.49% and 99.14% for 8[Formula: see text]s and (iii) 98.32%, 95.16% and 99.20% for 10[Formula: see text]s of ECG segments using only ten features. The integration of the proposed algorithm with ECG acquisition systems in the intensive care units (ICUs) can help the clinicians to decipher the shockable and nonshockable life-threatening arrhythmias accurately. Hence, doctors can use the CPR or AED immediately and increase the chance of survival during shockable life-threatening arrhythmia intervals.

A novel algorithm for ventricular arrhythmia classification using a fuzzy logic approach

In the present study, it has been shown that an unnecessary implantable cardioverter-defibrillator (ICD) shock is often delivered to patients with an ambiguous ECG rhythm in the overlap zone between ventricular tachycardia (VT) and ventricular fibrillation (VF); these shocks significantly increase mortality. Therefore, accurate classification of the arrhythmia into VT, organized VF (OVF) or disorganized VF (DVF) is crucial to assist ICDs to deliver appropriate therapy. A classification algorithm using a fuzzy logic classifier was developed for accurately classifying the arrhythmias into VT, OVF or DVF. Compared with other studies, our method aims to combine ten ECG detectors that are calculated in the time domain and the frequency domain in addition to different levels of complexity for detecting subtle structure differences between VT, OVF and DVF. The classification in the overlap zone between VT and VF is refined by this study to avoid ambiguous identification. The present method was trained and tested using public ECG signal databases. A two-level classification was performed to first detect VT with an accuracy of 92.6 %, and then the discrimination between OVF and DVF was detected with an accuracy of 84.5 %. The validation results indicate that the proposed method has superior performance in identifying the organization level between the three types of arrhythmias (VT, OVF and DVF) and is promising for improving the appropriate therapy choice and decreasing the possibility of sudden cardiac death.

Characterisation of the complexity of intracranial pressure signals measured from idiopathic and secondary normal pressure hydrocephalus patients

Hydrocephalus is a condition characterised by enlarged cerebral ventricles, which in turn affects intracranial pressure (ICP); however, the mechanisms regulating ICP are not fully understood. A nonlinear signal processing approach was applied to ICP signals measured during infusion studies from patients with two forms of hydrocephalus, in a bid to compare the differences. This is the first study of its kind. The two forms of hydrocephalus were idiopathic normal pressure hydrocephalus (iNPH) and secondary normal pressure hydrocephalus (SH). Following infusion tests, the Lempel-Ziv (LZ) complexity was calculated from the iNPH and SH ICP signals. The LZ complexity values were averaged for the baseline, infusion, plateau and recovery stages of the tests. It was found that as the ICP increased from basal levels, the LZ complexities decreased, reaching their lowest during the plateau stage. However, the complexities computed from the SH ICP signals decreased to a lesser extent when compared with the iNPH ICP signals. Furthermore, statistically significant differences were found between the plateau and recovery stage complexities when comparing the iNPH and SH results (p = 0.05). This Letter suggests that advanced signal processing of ICP signals with LZ complexity can help characterise different types of hydrocephalus in more detail.

Structured prediction for differentiating between normal rhythms, ventricular tachycardia, and ventricular fibrillation in the ECG

Recent studies have been performed on feature selection for diagnostics between non-ventricular rhythms and ventricular arrhythmias, or between non-ventricular fibrillation and ventricular fibrillation. However they did not assess classification directly between non-ventricular rhythms, ventricular tachycardia and ventricular fibrillation, which is important in both a clinical setting and preclinical drug discovery. In this study it is shown that in a direct multiclass setting, the selected features from these studies are not capable at differentiating between ventricular tachycardia and ventricular fibrillation. A high dimensional feature space, Fourier magnitude spectra, is proposed for classification, in combination with the structured prediction method conditional random fields. An improvement in overall accuracy, and sensitivity of every category under investigation is achieved.

Using Lempel-Ziv Complexity to Assess ECG Signal Quality

The poor quality of wireless electrocardiography (ECG) recordings can lead to misdiagnosis and waste of medical resources. This study presents an interpretation of Lempel–Ziv (LZ) complexity in terms of ECG quality assessment, and verifies its performance on real ECG signals. Firstly, LZ complexities for typical signals, namely high-frequency (HF) noise, low-frequency (LF) noise, power-line (PL) noise, impulse (IM) noise, clean artificial ECG signals, and ECG signals with various types of noise added (ECG plus HF, LF, PL, and IM noise, respectively) were analyzed. Then, the effects of noise, signal length, and signal-to-noise ratio (SNR) on the LZ complexity of ECG signals were analyzed. The simulation results show that LZ complexity for HF noise was obviously different from those for PL and LF noise. The LZ value can be used to determine the presence of HF noise. ECG plus HF noise had the highest LZ values. Other types of noise had low LZ values. Signal lengths of over 40 s had only a small effect on LZ values. The LZ values for ECG plus all types of noise increased monotonically with decreasing SNR except for LF and PL noise. For the test of real ECG signals plus three types of noise, namely muscle artefacts (MAs), baseline wander (BW), and electrode motion (EM) artefacts, LZ complexity varied obviously with increasing MA but not for BW and EM noise. This study demonstrates that LZ complexity is sensitive to noise level (especially for HF noise) and can thus be a valuable reference index for the assessment of ECG signal quality.

A novel encoding Lempel-Ziv complexity algorithm for quantifying the irregularity of physiological time series

BACKGROUND AND OBJECTIVE

The Lempel-Ziv (LZ) complexity and its variants have been extensively used to analyze the irregularity of physiological time series. To date, these measures cannot explicitly discern between the irregularity and the chaotic characteristics of physiological time series. Our study compared the performance of an encoding LZ (ELZ) complexity algorithm, a novel variant of the LZ complexity algorithm, with those of the classic LZ (CLZ) and multistate LZ (MLZ) complexity algorithms.

METHODS AND RESULTS

Simulation experiments on Gaussian noise, logistic chaotic, and periodic time series showed that only the ELZ algorithm monotonically declined with the reduction in irregularity in time series, whereas the CLZ and MLZ approaches yielded overlapped values for chaotic time series and time series mixed with Gaussian noise, demonstrating the accuracy of the proposed ELZ algorithm in capturing the irregularity, rather than the complexity, of physiological time series. In addition, the effect of sequence length on the ELZ algorithm was more stable compared with those on CLZ and MLZ, especially when the sequence length was longer than 300. A sensitivity analysis for all three LZ algorithms revealed that both the MLZ and the ELZ algorithms could respond to the change in time sequences, whereas the CLZ approach could not. Cardiac interbeat (RR) interval time series from the MIT-BIH database were also evaluated, and the results showed that the ELZ algorithm could accurately measure the inherent irregularity of the RR interval time series, as indicated by lower LZ values yielded from a congestive heart failure group versus those yielded from a normal sinus rhythm group (p < 0.01).

Efficient and robust ventricular tachycardia and fibrillation detection method for wearable cardiac health monitoring devices

In this Letter, the authors propose an efficient and robust method for automatically determining the VT and VF events in the electrocardiogram (ECG) signal. The proposed method consists of: (i) discrete cosine transform (DCT)-based noise suppression; (ii) addition of bipolar sequence of amplitudes with alternating polarity; (iii) zero-crossing rate (ZCR) estimation-based VTVF detection; and (iv) peak-to-peak interval (PPI) feature based VT/VF discrimination. The proposed method is evaluated using 18,000 episodes of different ECG arrhythmias taken from 6 PhysioNet databases. The method achieves an average sensitivity (Se) of 99.61%, specificity (Sp) of 99.96%, and overall accuracy (OA) of 99.92% in detecting VTVF and non-VTVF episodes by using a ZCR feature. Results show that the method achieves a Se of 100%, Sp of 99.70% and OA of 99.85% for discriminating VT from VF episodes using PPI features extracted from the processed signal. The robustness of the method is tested using different kinds of ECG beats and various types of noises including the baseline wanders, powerline interference and muscle artefacts. Results demonstrate that the proposed method with the ZCR, PPI features can achieve significantly better detection rates as compared with the existing methods.

Reducing false arrhythmia alarms in the ICU using multimodal signals and robust QRS detection

This study developed algorithms to decrease the arrhythmia false alarms in the ICU by processing multimodal signals of photoplethysmography (PPG), arterial blood pressure (ABP), and two ECG signals. The goal was to detect the five critical arrhythmias comprising asystole (ASY), extreme bradycardia (EBR), extreme tachycardia (ETC), ventricular tachycardia (VTA), and ventricular flutter or fibrillation (VFB). The different characteristics of the arrhythmias suggested the application of individual signal processing for each alarm and the combination of the algorithms to enhance false alarm detection. Thus, different features and signal processing techniques were used for each arrhythmia type. The ECG signals were first processed to reduce the signal interference. Then, a Hilbert-transform based QRS detector algorithm was utilized to identify the QRS complexes, which were then processed to determine the instantaneous heart rate. The pulsatile signals (PPG and ABP) were processed to discover the pulse onset of beats which were then employed to measure the heart rate. The signal quality index (SQI) of the signals was implemented to verify the integrity of the heart rate information. The overall score obtained by our algorithms in the 2015 Computing in Cardiology Challenge was a score of 74.03% for retrospective and 69.92% for real-time analysis.