Narrow Results

The repolarization and depolarization in heart generate electrical signals in the form of an ECG wave. The condition of the heart can be indicated by using Heart Rate Variability (HRV) features. In this work, FIR filter is used at the pre-processing phase for denoising, and then statistical analysis is applied for time-domain HRV feature extraction and selection. This algorithm is evaluated on different records of MIT/BIH Normal Sinus Rhythm and Arrhythmia database. The $t$-test implementation in both databases shows that there are significant variations in HRV features, where meanRR and HR have suggestive significant ($0.05<p\le 0.10$) changes, while maxRR, minRR, maxminRR, and SDNN have strongly significant ($p\le 0.01$) changes. To validate the statistical analysis of HRV, feature classification has been done using SVM and kNN classifiers. A significant improvement of 2% and 14.02% has been observed in the overall accuracy of SVM and kNN classifiers after feature selection, respectively. These HRV features can be used for the early prediction of various Cardio-Vascular Diseases (CVD).

The heart beat RR intervals extracted from the electrocardiogram recorded during the stress test show a non stationary profile consisting of a decreasing trend during the exercise phase, an increasing trend during the recovery and a global minimum (acme). In addition this time series exhibits a time-varying variance. We decompose the series into a deterministic trend and random fluctuation. The trend is obtained as an exponential fit of the data; the fluctuation is modeled as a mean reverting process driven by the trend, in which the random innovation has a time-varying variance. Data analysis, performed on ambulatory recorded electrocardiograms of 10 healthy subjects, shows that the model describes correctly the data series on a scale of at least 300 beats.

The RR and RT time intervals extracted from the electrocardiogram measure respectively the duration of cardiac cycle and repolarization. The series of these intervals recorded during the exercise test are characterized by two trends: A decreasing one during the stress phase and an increasing one during the recovery, separated by a global minimum. We model these series as a sum of a deterministic trend and random fluctuations, and estimate the trend using methods of curve extraction: Running mean, polynomial fit, multi scale wavelet decomposition. We estimate the minimum location from the trend. Data analysis performed on a group of 20 healthy subjects provides evidence that the minimum of the RR series precedes the minimum of the RT series, with a time delay of about 19 seconds.

In recent years, the number of cardiac disease patients has been increasing. Modern medical research has shown that the complexity of electrocardiogram (ECG) signals is related to cardiovascular diseases. This paper investigates the difference in complexity of ECG data from the people with different cardiovascular diseases, such as atrial fibrillation (AF), ventricular arrhythmia (VA) and congestive heart failure (CHF). The empirical mode decomposition (EMD) and multiscale entropy method are used to analyze the ECG data, and a mathematical model established by a support vector machine is used to identify different diseases. The accuracy recognition rate of the AF recognition is 96.25%, and that of the CHF and VA reach 90.26% and 92.20%, respectively. The experimental results show that the recognition method proposed in this paper is successful.

This quantitative study identifies the coupling changes occurring among cardiac (RR), vascular (SBP) and respiratory (RESP) signals during deep breathing. The deep breathing measures the dysfunction of the parasympathetic autonomic nervous system. The traditional methods based on cross-correlation and coherence analysis lack to measure nonlinear structures and unpredictability of physiological subsystems. Therefore, information domain coupling method based on conditional entropy is proposed to detect the coupling changes. Thirty healthy volunteers were examined for 5min at normal breathing and 5min during deep breathing (6cycles/min). The reduction in respiration rate detects a significant increase in information flow from RESP to RR, RESP to SBP and SBP to RR. The increased interaction from RESP to RR and RESP to SBP at reduced respiration rate indicates the enhancement of respiratory sinus arrhythmia that results in the activation of the parasympathetic nervous system. Also, the balanced cardiovascular interaction observed on normal breathing from RR to SBP disappears, but interaction occurring in baroreflex direction (SBP to RR) increases that helps in the reduction of blood pressure during deep breathing. This detected direction of information flow helps in identifying the coupling changes occurring during parasympathetic nerve activity.

Many physiological signals such as heart rate (HR), blood pressure (BP), and respiration (RESP) affect each other, and the inter-relation within and between these signals can be linear or nonlinear. Therefore, this paper’s main aim is to extract the relevant features using the information domain coupling technique based on conditional transfer entropy to detect the nonlinearity and coupling changes between the physiological signals and to classify the database using various machine learning classifiers to study the aging changes in the contribution of HR, BP, and RESP. In the proposed work, the physiological signals, i.e. HR, BP, and RESP, were pre-processed using various filtering methods, then features of physiological signals were extracted using linear and nonlinear techniques. After the pre-processing and extraction of features, the extracted features are classified using machine learning classifiers to classify the physiological signal database to study the aging changes in the contribution of HR, BP, and RESP. The data has been taken from the standard Fantasia database of healthy young and old subjects and self-recorded data of healthy young and old subjects for this study. Naive Bayes (NB), Support vector machine (SVM), K-Nearest Neighbor (KNN), Logistic Regression (LR), and Artificial Neural Network (ANN) were trained using five-fold cross-validation on the physiological dataset. It is concluded from the results that by adding the coupling features, the efficiency of the final prediction of the classifier increased from $70$% to $75$% obtained by LR, $90$% to $100$% obtained by SVM, $95$% to $100$% obtained by KNN, $81.2$% to $85$% obtained by NB, and $95$% to $100$% obtained by ANN. The ANN performs well when provided with the coupling features, gives a maximum accuracy of $100$% and very high sensitivity of $100$% and specificity of $100$%, and takes much less computational time, when compared to other machine learning algorithms on same length of database.