The influences of nonstationarity and nonlinearity on heart rate time series can be mathematically qualified or quantified by multiscale entropy (MSE). The aim of this study is to investigate the prognostic value of parameters derived from MSE in the patients with systolic heart failure. Patients with systolic heart failure were enrolled in this study. One month after clinical condition being stable, 24-hour Holter electrocardiogram was recording. MSE as well as other standard parameters of heart rate variability (HRV) and detrended fluctuation analysis (DFA) were assessed. A total of 40 heart failure patients with a mea age of 56±16 years were enrolled and followed-up for 684±441 days. There were 25 patients receiving β-blockers treatment. During follow-up period, 6 patients died or received urgent heart transplantation. The short-term exponent of DFA and the slope of MSE between scale 1 to 5 were significantly different between patients with or without β-blockers (p = 0.014 and p = 0.028). Only the area under the MSE curve for scale 6 to 20 (Area_6–20) showed the strongest predictive power between survival (n = 34) and mortality (n = 6) groups among all the parameters. The value of Area_6–20 21.2 served as a significant predictor of mortality or heart transplant (p = 0.0014). The area under the MSE curve for scale 6 to 20 is not relevant to β-blockers and could further warrant independent risk stratification for the prognosis of CHF patients.

Advanced deep learning models have shown excellent performance in the task of fault bearing diagnosis over the traditional machine learning and signal processing techniques. Few-shot learning approach has also been attracting a lot of attention in this task to address the problem of limited training data. Nevertheless, cutting-edge models for fault bearing diagnosis are often based on convolutional neural networks that emphasize on local features of input data. Besides, accurate classification of fault bearing signals is still nontrivial due to the variations of data, fault types, acquisition conditions, and extremely limited data, leaving a space for research on this topic. In this study, we propose a novel end-to-end approach for fault bearing diagnosis even in the case of limited data with artificial and real faults. In particular, we propose a module for automatic feature extraction from input data namely Multiscale Large Kernel Feature Extraction. The extracted features are then fed into a two-branch model including a global and a local branch. The global one includes a transformer architecture with cross-attention to handle global context and obtain the correlation between the query and support sets. The local branch is a metric-based model consisting of Mahalanobis distance for separating local features from the support set. The outputs from the two branches are then ensembled for classification purpose. Intensive experiments and ablation studies have been made on the two public datasets including CWRU and PU. Qualitative and quantitative results with different degrees of training samples by the proposed model in comparison with other state-of-the-arts have shown the superior performance of the proposed approach. Our code will be published at: https://github.com/HungVu307/Few-shot-via-ensembling-Transformer-with-Mahalanobis-distance.

The empirical mode decomposition (EMD) is an established method for the time-frequency analysis of nonlinear and nonstationary signals. However, one major drawback of the EMD is the mode mixing effect. Many modifications have been made to resolve the mode mixing effect. In particular, disturbance-assisted EMDs, such as the noise-assisted EMD and the masking EMD, have been proposed to resolve this problem. These disturbance-assisted approaches have led to a better performance of the EMD in the analysis of real-world data sets, but they may also have two side effects: the mode splitting and residual noise effects. To minimize or eliminate the mode mixing effect while avoiding the two side effects of traditional disturbance-assisted EMDs, we propose an EMD-based algorithm assisted by sinusoidal functions with a designed uniform phase distribution with a comprehensive theoretical explanation for the substantial reduction of the mode splitting and the residual noise effects simultaneously. We examine the performance of the new method and compare it to those of other disturbance-assisted EMDs using synthetic signals. Finally numerical experiments with real-world examples are conducted to verify the performance of the proposed method.

Monitoring fetal heart rate during pregnancy is essential to assist clinicians in making more timely decisions. Non-invasive monitoring of fetal heart activities using abdominal ECGs is useful for diagnosis of heart defects. However, the extracted fetal ECGs are usually too weak to be robustly detected. Thus, it is a necessity to enhance fetal R-peak since their peaks may hidden within the signal due to the immaturity of the fetal cardiovascular system. Therefore, to improve the detection of the fetal heartbeat, a novel fetal R-peak enhancement technique was proposed to statistically generate the weighting mask according to the distribution of the neighboring temporal intervals between each pair of peaks. Two set of simulations were designed to validate the reliability of the method: challenges with different levels of (1) noise contamination and (2) R-peak interval changing rate. The simulation results showed that the weighting mask improve the accuracy of the R-peak detection rate by 25% and decrease false alarm rate by 20% with white noise contamination, and ensure high R-peak detection rate (> 80%), especially with mild noise contamination (noise amplitude ratio < 1.5 and noise rate per minute < 25%). For those simulations with continuous R-peak intervals changing, the masking process can still effectively eliminate noise contamination especially when the amplitude of the sinusoidal fetal R-R intervals lower than 50msec. For the real fetus ECGs, detection rate was increasing by 3.498% whereas false alarm rate was decreasing by 3.933%. Next, we implement the fetal R-peak enhancement technique to investigate fractal regulation and multiscale entropy of the real fetal heartbeat intervals. Both scaling exponent (∼0.6 to ∼1 in scale 4-15) and entropy measure (scale 6-10) increase from with gestational ages (22-40 weeks). The results confirmed fractal slope and complexity of fetal heartbeat intervals can reflect the maturation of fetus organism.