Channel ECG Research Articles

Automatic Detection of Occluded Main Coronary Arteries of NSTEMI Patients with MI-MS ConvMixer + WSSE Without CAG

Background/Objectives: Heart attacks are the leading cause of death in the world. There are two important classes of heart attack: ST-segment Elevation Myocardial Infarction (STEMI) and Non-ST-segment Elevation Myocardial Infarction (NSTEMI) patient groups. While the STEMI group has a higher mortality rate in the short term, the NSTEMI group is considered more dangerous and insidious in the long term. Blocked coronary arteries can be predicted from ECG signals in STEMI patients but not in NSTEMI patients. Therefore, coronary angiography (CAG) is inevitable for these patients. However, in the elderly and some patients with chronic diseases, if there is a single blockage, the CAG procedure poses a risk, so medication may be preferred. In this study, a novel deep learning-based approach is used to automatically detect the occluded main coronary artery or arteries in NSTEMI patients. For this purpose, a new seven-class dataset was created with expert cardiologists. Methods: A new Multi Input-Multi Scale (MI-MS) ConvMixer model was developed for automatic detection. The MI-MS ConvMixer model allows simultaneous training of 12-channel ECG data and highlights different regions of the data at different scales. In addition, the ConMixer structure provides high classification performance without increasing the complexity of the model. Moreover, to maximise the classifier performance, the WSSE algorithm was developed to adjust the classification prediction value according to the feature importance weights. Results: This algorithm improves the SVM classifier performance. The features extracted from this model were classified with the WSSE algorithm, and an accuracy of 88.72% was achieved. Conclusions: This study demonstrates the potential of the MI-MS ConvMixer model in advancing ECG signal classification for diagnosing coronary artery diseases, offering a promising tool for real-time, automated analysis in clinical settings. The findings highlight the model’s ability to achieve high sensitivity, specificity, and precision, which could significantly improve.

The ECG of a 6-year-old girl

A 12-channel ECG and the ECG patterb from Holter monitoring of a 6-year-girl are presented.

MDDBranchNet: A Deep Learning Model for Detecting Major Depressive Disorder Using ECG Signal.

Major depressive disorder (MDD) is a chronic mental illness which affects people's well-being and is often detected at a later stage of depression with a likelihood of suicidal ideation. Early detection of MDD is thus necessary to reduce the impact, however, it requires monitoring vitals in daily living conditions. EEG is generally multi-channel and due to difficulty in signal acquisition, it is unsuitable for home-based monitoring, whereas, wearable sensors can collect single-channel ECG. Classical machine-learning based MDD detection studies commonly use various heart rate variability features. Feature generation, which requires domain knowledge, is often challenging, and requires computation power, often unsuitable for real time processing, MDDBranchNet is a proposed parallel-branch deep learning model for MDD binary classification from a single channel ECG which uses additional ECG-derived signals such as R-R signal and degree distribution time series of horizontal visibility graph. The use of derived branches was able to increase the model's accuracy by around 7%. An optimal 20-second overlapped segmentation of ECG recording was found to be beneficial with a 70% prediction threshold for maximum MDD detection with a minimum false positive rate. The proposed model evaluated MDD prediction from signal excerpts, irrespective of location (first, middle or last one-third of the recording), instead of considering the entire ECG signal with minimal performance variation stressing the idea that MDD phenomena are likely to manifest uniformly throughout the recording.

PCA-Based Preprocessing for Clustering-Based Fetal Heart Rate Extraction in Non-Invasive Fetal Electrocardiograms

Non-invasive fetal electrocardiography (NI-ECG) is based on the acquisition of signals from electrodes on the mother’s abdominal surface. This abdominal ECG (aECG) signal consists of the maternal ECG (mECG) along with the fetal ECG (fECG) and other noises and artifacts. These records allow the acquisition of valuable and reliable information that helps ensure fetal well-being during pregnancy. This paper proposes a procedure based on principal component analysis (PCA) to obtain a single-channel master abdominal ECG record that can be used as input to fetal heart rate extraction techniques. The new procedure requires three main processing stages: PCA-based analysis for fECG-component extraction, polarity test, and curve fitting. To show the advantages of the proposal, this PCA-based method has been used as the feeding stage to a previously developed clustering-based method for single-channel aECG fetal heart rate monitoring. The results obtained for a set of real abdominal ECG recordings from annotated public aECG databases, the Abdominal and Direct Fetal ECG Database and the Challenge 2013 Training Set A, show improved efficiency in fetal heart rate extraction and illustrate the benefits derived from the use of such a master abdominal ECG channel. This allows us to achieve proper fetal heart rate monitoring without the need for manual inspection and selection of channels to be processed, while also allowing us to analyze records that would have been discarded otherwise.

Predictors of Atrial Fibrillation in Patients with Embolic Stroke of Unknown Etiology and Implantable Loop Recorders-Further Insights of the TRACK AF Study on the Role of ECG and Echocardiography.

Aims-Electrocardiography (ECG) and echocardiographic left atrial (LA) parameters may be helpful to assess the risk of atrial fibrillation (AF) in embolic stroke of unknown etiology (ESUS) and could therefore guide intensity of ECG monitoring. Methods-1153 consecutive patients with ischemic stroke or transient ischemic attack (TIA) were analyzed. An internal loop recorder (ILR) was implanted in 104 consecutive patients with ESUS. Multiple morphologic P-wave parameters in baseline 12-channel ECG and echocardiographic LA parameters were measured and analyzed in patients with and without ILR-detected AF. Using logistic regression, we evaluated the predictive value of several ECG parameters and LA dimensions on the occurrence of AF. Results-In 20 of 104 (19%) patients, AF was diagnosed by ILR during a mean monitoring time of 575 (IQR 470-580) days. Patients with AF were significantly older (72 (67-75) vs. 60 (52-72) years; p = 0.001) and premature atrial contractions (PAC) were more frequently observed (40% vs. 2%; p < 0.001) during baseline ECG. All morphologic P-wave parameters did not show a significant difference between groups. There was a non-significant trend towards a larger LA volume index (31 (24-36) vs. 29 (25-37) mL/m2; p = 0.09) in AF patients. Conclusions-Age and PAC are independently associated with incident AF in ESUS and could be used as markers for selecting patients that may benefit from more extensive rhythm monitoring or ILR implantation. In our consecutive cohort of patients with ESUS, neither morphological P-wave parameters nor LA size were predictive of AF.

Ultra-High-Resolution Electrocardiography Enables Earlier Detection of Transmural and Subendocardial Myocardial Ischemia Compared to Conventional Electrocardiography.

The sensitivity of exercise ECG is marginally sufficient for the detection of mild reduction of coronary blood flow in patients with early coronary atherosclerosis. Here, we describe the application of a new technique of ECG registration/analysis-ultra-high-resolution ECG (UHR ECG)-for early detection of myocardial ischemia (MIS). The utility of UHR ECG vs. conventional ECG (C ECG) was tested in anesthetized rats and pigs. Transmural MIS was induced in rats by the ligation of the left coronary artery (CA). In pigs, subendocardial ischemia of a variable extent was produced by stepwise inflation of a balloon within the right CA, causing a 25-100% reduction of its lumen. In rats, a reduction in power spectral density (PSD) in the high-frequency (HF) channel of UHR ECG was registered at 60 s after ischemia (power 0.81 ± 0.14 vs. 1.25 ± 0.12 mW at baseline, p < 0.01). This was not accompanied by any ST segment elevation on C ECG. In pigs, PSD in the HF channel of UHR ECG was significantly decreased at a 25% reduction of CA lumen, while the ST segment on C ECG remained unchanged. In conclusion, UHR ECG enabled earlier detection of transmural MIS compared to C ECG. PSD in the HF channel of UHR ECG demonstrated greater sensitivity in the settings of subendocardial ischemia.

MACHINE LEARNING APPROACH TO DETECT ECG ABNORMALITIES USING COST-SENSITIVE DECISION TREE CLASSIFIER

Cardiac Arrhythmia is an abnormal heart rhythm that develops when the electrical impulses control the heart’s contraction which does not function properly. The heart can beat too fast (tachycardia), too slow (bradycardia), or in an irregular pattern. Observing ECG signal peaks and channels freehand is difficult due to their ingenious modification. Automated detection of cardiovascular abnormalities is preferred for the early diagnosis of cardiac disorders. This paper used machine learning approaches for detecting ECG abnormality utilizing a Support Vector Machine (SVM) and Cost-Sensitive Decision-Tree (CS-DT) classifier. The Empirical Mode Decomposition approach was utilized to examine the properties of R-peaks and QRS complexes in ECG signs. Various morphological characteristics are analyzed from the signal penetrated by the classifier to diagnose the irregular beats. A set of twenty-two clinically feasible features comprising temporal, morphological, and statistical were extracted from the processed ECG signals and applied to the classifier to categorize cardiovascular irregularities like Normal (N), Left Bundle Branch Block (LBBB), Right Bundle Branch Block (RBBB), Atrial Premature Beats (APB), and Premature Ventricular Contraction (PVC). The Beth Israel Hospital at Massachusetts Institute of Technology (MIT-BIH) dataset has been used for this work, where feature datasets are split into training and evaluation subsets. The training set is used to train machine learning models on the extracted features, while the evaluation set is used to assess the performance of the trained models. The evaluation metrics such as Accuracy (Acc), Sensitivity (Se), Specificity (Sp), and Positive Predictivity (Pp), are frequently used to evaluate the model’s performance in Arrhythmia detection along with classification. The simulation has been conducted using SVM and CS-DT classifier with performance for all individual class labels at a Confidence Factor (CF) of 0.5. The performance of the time and frequency domain features is merged resulting in higher classification of Sensitivity, Specificity, Positive Predictivity, and Accuracy of 89.5%, 98.11%, 87.76%, and 96.8% in SVM, 97.71%, 99.58%, 97.66%, 99.32% in CS-DT classifier in identifying the irregular heartbeats.

ECG Changes in Patients with Amitriptyline Intoxication; P, QT, QTC, T and PR Wave Dispersion: A Retrospective Study

Objective: Amitriptyline intoxication is a life-threatening condition as it can cause cardiac arrhythmias. This study aims to examine the ECG changes in patients with amitriptyline intoxications who were admitted to the emergency department and to compare the P-wave, T-wave, QT, QTC, and PR dispersions with healthy controls.&#x0D; Material and Methods: The sample of the current retrospective study consists of only 46 patients with amitriptyline intoxication and 65 healthy controls. The 12-channel ECG recordings of the participants were evaluated by two experienced cardiologists. P wave, T wave, QT, QTC, and PR distance, minimum, maximum, and dispersion values were measured. The obtained data were recorded on the data set and subjected to statistical analysis.&#x0D; Results: There was no significant difference in sociodemographic variables between patients with amitriptyline intoxication and the control group. The main symptoms of the patients were as follows: 28.3% (n=13) lethargy, 32.6% (n=15) nausea-vomiting, 21.7% (n=10) unconsciousness, and 17.4% (n=8) other. According to the analysis of amitriptyline intoxication and ECG parameters of healthy controls, P-wave, T-wave, QT, QTC, and PR dispersion were higher in the intoxication group than in the healthy controls (p0.05). The patient group had a higher heart rate than the control group (p=0.026).&#x0D; Conclusion: The current study found that P-wave, T-wave, QT, QTC, and PR dispersion were higher in the amitriptyline intoxication group than in the control group, which is thought to play a role in cardiac arrhythmias. Clinicians should be on alert for any ECG changes in these cases.

Self-Supervised Representation Learning-Based OSA Detection Method Using Single-Channel ECG Signals

Sleep apnea(SA) is a pervasive and highly prevalent sleep disorder identified by recurrent breathing-related problems such as respiratory pauses for almost 10 seconds (called apnea events) during sleep. It is a strongly underdiagnosed problem because the person suffering from this disease is not aware of this situation. It may cause serious health issues and badly affect the quality of life. Therefore, the diagnosis of sleep is crucial to cure disease. Polysomnography(PSG) is a golden technique for diagnosing sleep disorders. In this technique, multiple sensors are used to collect specific physiological signals like Electroencephalogram(EEG), Electromyogram(EMG), electrooculogram(EOG), and many more. In regular clinical practice, medical experts need to manually analyze the signals of sleep hours which is a tedious process. Therefore, the automatic diagnosis tool is needed to simplify this process. Recently, many research groups have proposed deep-learning models for the automatic diagnosis of sleep apnea using physiological signals with good accuracy. However, all these models require a large amount of annotated data in the supervised training process, which limits the use of those models in real-time scenarios. However, annotating a huge amount of biomedical signals is challenging and requires lots of time and domain expertise. This study proposes a self-supervised representation learning method for detecting hypopnea events from single channel ECG signals. The proposed model is trained in two phases. In the first training phase, an encoder is trained to learn signal representation from the unlabeled data. In the second training phase, the classifier and the encoder are finetuned for the classification. Our proposed model performed well on the test data set with a per-segment classification accuracy of 85%, 89% and 92% by using only 1%, 10%, and 100% of training data with labels, respectively, for fine-tuning encoder along with the classifier. Also, our proposed model can identify a person suffering from the OSA with the accuracy of 100%, even when the encoder and classifier are fine-tuned using only 1% of training data with the label. The proposed model outperformed the state-of-the-art techniques and can be implemented offline or online for rapid and accurate diagnosis of the problem.

Review of current ECG consumer electronics (pros and cons)

BackgroundSeveral wearable, medical-grade consumer ECG devices are now available and integrated into consumer electronics like multi sensor fitness watches and scales. Specific consumer ECGs can also come in the form of patches or thin sensor plates in credit card or other shapes. Watches with ECG capabilities are often multi vital sign sensor devices. The majority of these devices are usually connected to a mobile smartphone. However, there are pros and cons to their use. MethodsWe review here an exemplary selection of modern consumer ECG devices based on device type, recording method and the number of standard ECG channels derived. ResultsSingle-channel consumer ECG devices such as Smart Watches can be useful for detecting and monitoring atrial fibrillation and flutter and other arrhythmias, as well as ectopic complexes. However, they are currently limited with respect to recording duration and information content (a single-channel or limb‑lead ECG having less diagnostic information than a 12‑lead ECG). While some non watch-based consumer ECG devices can now record all 6 limb leads to yield increased information, no consumer ECG devices can currently reliably detect ST-segment deviations, potentially indicating myocardial infarction or ischemic episodes. Moreover, barriers to use still exist for at-risk elderly people. Finally, there currently is no universal data exchange format. ConclusionConsumer ECG devices, whether in fitness or fashionable design, allow for reliable detection of atrial fibrillation. Timely detection of atrial fibrillation and subsequent treatment might protect against stroke, especially in high-risk groups, yet prospective evidence is still lacking. Six-channel consumer ECG and longer data collection capabilities extend potential functionality, including for the monitoring of ST-segments and QT intervals. However, no currently available devices are sufficiently suitable for the detection of myocardial infarction or ischemia, which is why portable 12-channel technologies are desirable. For the reliable detection of a myocardial infarction, the determination of specific myocardial infarction blood markers and evaluation of patient medical history still is indispensable in addition to the 12 lead ECG.

Method to quantify the impact of sleep on cardiac neuroautonomic functionality: application to the prediction of cardiovascular events in the Multi-Ethnic Study of Atherosclerosis.

We introduce the concept of cardiac neuroautonomic renewability and a method for its quantification. This concept refers to the involuntary nervous system's capacity to improve cardiac control in response to restorative interventions, such as sleep. We used the change in heart rate fragmentation (ΔHRF), before sleep onset compared with after sleep termination, to quantify the restorative effects of sleep. We hypothesized that the ability to improve cardiac neuroautonomic functionality would diminish with age and be associated with lower risk of major adverse cardiovascular events (MACE). We analyzed the ECG channel of polysomnographic recordings from an ancillary investigation of the Multi-Ethnic Study of Atherosclerosis (MESA). In a cohort of 659 participants (mean ± SD age, 69.7 ± 8.8; 42% male), HRF was significantly (P < 0.001) lower after sleep (before: 74 ± 12%, after: 67 ± 13%). Furthermore, the magnitude of the decrease significantly (P < 0.001) diminished with cross-sectional age. In addition, a larger reduction in HRF following sleep (i.e., higher ΔHRF) was associated with lower risk of MACE, independent of traditional cardiovascular risk factors and current measures of sleep quality. Specifically, over a mean follow-up period of 6.4 ± 1.6 yr, in which 60 participants had their first MACE, a one-SD (12%) increase in ΔHRF was associated with a 36% (95% CI: 12%-53%) decrease in the risk of MACE. The results demonstrate the restorative impact of sleep on heart rate control. As such they support the concept of cardiac neuroautonomic renewability and the utility of ΔHRF for its quantification.

Diagnostic Accuracy of a Portable ECG Device in Rowing Athletes.

Background: Athletes can experience exercise-induced transient arrythmias during high-intensity exercise or competition, which are difficult to capture on traditional Holter monitors or replicate in clinical exercise testing. The aim of this study was to investigate the reliability of a portable single channel ECG sensor and data recorder (PluxECG) and to evaluate the confidence and reliability in interpretation of ECGs recorded using the PluxECG during remote rowing. Methods: This was a two-phase study on rowing athletes. Phase I assessed the accuracy and precision of heart rate (HR) using the PluxECG system compared to a reference 12-lead ECG system. Phase II evaluated the confidence and reliability in interpretation of ECGs during ergometer (ERG) and on-water (OW) rowing at moderate and high intensities. ECGs were reviewed by two expert readers for HR, rhythm, artifact and confidence in interpretation. Results: Findings from Phase I found that 91.9% of samples were within the 95% confidence interval for the instantaneous value of the changing exercising HR. The mean correlation coefficient across participants and tests was 0.9886 (σ = 0.0002, SD = 0.017) and between the two systems at elevated HR was 0.9676 (σ = 0.002, SD = 0.05). Findings from Phase II found significant differences for the presence of artifacts and confidence in interpretation in ECGs between readers’ for both intensities and testing conditions. Interpretation of ECGs for OW rowing had a lower level of reader agreement than ERG rowing for HR, rhythm, and artifact. Using consensus data between readers’ significant differences were apparent between OW and ERG rowing at high-intensity rowing for HR (p = 0.05) and artifact (p = 0.01). ECGs were deemed of moderate-low quality based on confidence in interpretation and the presence of artifacts. Conclusions: The PluxECG device records accurate and reliable HR but not ECG data during exercise in rowers. The quality of ECG tracing derived from the PluxECG device is moderate-low, therefore the confidence in ECG interpretation using the PluxECG device when recorded on open water is inadequate at this time.

Fitness, performance, and cardiac autonomic responses to exercise in people with epilepsy

People with epilepsy (PWE) are less fit and have an increased risk of sudden cardiac death. Imbalances within the autonomic nervous system (ANS) are believed to mediate some of those effects. However, results are mostly derived from patients whose seizures are refractory to medical therapy. In this study, an exhaustive bicycle ergometer test was delivered to 25 PWE (19 seizure free in the last 6 months) recruited in a community-based setting and 25 age-, sex-, and BMI-matched healthy controls. During the exercise test a 12-channel ECG was recorded and spirometry was carried out to determine the maximal oxygen uptake (VO2peak) as the gold standard to assess fitness. Before and after exercise, heart rate variability (HRV) and electrodermal activity (EDA) were measured along with an electroencephalogram (EEG). Blood samples were collected to determine anti-seizure drug (ASD) serum levels and physical activity of daily living was evaluated via the International Physical Activity Questionnaire (IPAQ). People with epilepsy and healthy controls were similarly fit and physically active. However, PWE had a lower maximum heart rate, a lower heart rate reserve, and a lower chronotropic index. The ratio between low- to high-frequency HRV changes (LF/HF ratio) was lower in PWE. Two patients with idiopathic genetic epilepsies revealed generalized interictal epileptiform discharges only after, but not before exercise. However, post-exercise EEG measurement was three times longer than pre-exercise and those patients did not report exercise induced seizures in the history. Besides epileptogenesis, anti-seizure medications may also contribute to those autonomic differences.

Use of Deep Learning to Detect the Maternal Heart Rate and False Signals on Fetal Heart Rate Recordings.

We have developed deep learning models for automatic identification of the maternal heart rate (MHR) and, more generally, false signals (FSs) on fetal heart rate (FHR) recordings. The models can be used to preprocess FHR data prior to automated analysis or as a clinical alert system to assist the practitioner. Three models were developed and used to detect (i) FSs on the MHR channel (the FSMHR model), (ii) the MHR and FSs on the Doppler FHR sensor (the FSDop model), and (iii) FSs on the scalp ECG channel (the FSScalp model). The FSDop model was the most useful because FSs are far more frequent on the Doppler FHR channel. All three models were based on a multilayer, symmetric, GRU, and were trained on data recorded during the first and second stages of delivery. The FSMHR and FSDop models were also trained on antepartum recordings. The training dataset contained 1030 expert-annotated periods (mean duration: 36 min) from 635 recordings. In an initial evaluation of routine clinical practice, 30 fully annotated recordings for each sensor type (mean duration: 5 h for MHR and Doppler sensors, and 3 h for the scalp ECG sensor) were analyzed. The sensitivity, positive predictive value (PPV) and accuracy were respectively 62.20%, 87.1% and 99.90% for the FSMHR model, 93.1%, 95.6% and 99.68% for the FSDop model, and 44.6%, 87.2% and 99.93% for the FSScalp model. We built a second test dataset with a more solid ground truth by selecting 45 periods (lasting 20 min, on average) on which the Doppler FHR and scalp ECG signals were recorded simultaneously. Using scalp ECG data, the experts estimated the true FHR value more reliably and thus annotated the Doppler FHR channel more precisely. The models achieved a sensitivity of 53.3%, a PPV of 62.4%, and an accuracy of 97.29%. In comparison, two experts (blinded to the scalp ECG data) respectively achieved a sensitivity of 15.7%, a PPV of 74.3%, and an accuracy of 96.91% and a sensitivity of 60.7%, a PPV of 83.5% and an accuracy of 98.24%. Hence, the models performed at expert level (better than one expert and worse than the other), although a well-trained expert with good knowledge of FSs could probably do better in some cases. The models and datasets have been included in the Fetal Heart Rate Morphological Analysis open-source MATLAB toolbox and can be used freely for research purposes.

The Effect of Aquatic Exercise Training on Heart Rate Variability in Patients with Coronary Artery Disease.

(1) Background: Aquatic exercise training is a relatively understudied exercise modality in patients with CAD; with the present study, we sought to compare the impact of short-term 14-day water- and land-based exercise training on heart rate variability (HRV). (2) Methods: We randomized 90 patients after a recent CAD event (myocardial infarction and/or revascularization within 2 months prior to inclusion) to either (i) water-based or (ii) land-based exercise training (14 days, two 30 min sessions daily), or (iii) controls. Before and after the intervention period, all participants underwent 20 min 12-channel high-resolution ECG recordings with off-line HRV analysis, including conventional linear time- and frequency-domain analysis (using the Welch method for fast-Fourier transformation), and preselected non-linear analysis (Poincaré plot-derived parameters, sample entropy, and the short-term scaling exponent α1 obtained by detrended fluctuation analysis). (3) Results: Eighty-nine patients completed the study (mean age 60 ± 8 years; 20 % women). We did not detect significant differences in baseline- or age-adjusted end-of-study HRV parameters, but aquatic exercise training was associated with a significant increase in the linear LF/HF parameter (from 2.6 [1.2–4.0] to 3.0 [2.1–5.5], p = 0.046) and the non-linear α1 parameter (from 1.2 [1.1–1.4] to 1.3 [1.2–1.5], p = 0.043). (4) Conclusions: Our results have shown that a short-term 14-day aquatic exercise training program improves selected HRV parameters, suggesting this mode of exercise is safe and may be beneficial in patients with CAD.

Robust R-Peak Detection in Low-Quality Holter ECGs Using 1D Convolutional Neural Network.

Noise and low quality of ECG signals acquired from Holter or wearable devices deteriorate the accuracy and robustness of R-peak detection algorithms. This paper presents a generic and robust system for R-peak detection in Holter ECG signals. While many proposed algorithms have successfully addressed the problem of ECG R-peak detection, there is still a notable gap in the performance of these detectors on such low-quality ECG records. In this study, a novel implementation of the 1D Convolutional Neural Network (CNN) is used integrated with a verification model to reduce the number of false alarms. This CNN architecture consists of an encoder block and a corresponding decoder block followed by a sample-wise classification layer to construct the 1D segmentation map of R-peaks from the input ECG signal. Once the proposed model has been trained, it can solely be used to detect R-peaks possibly in a single channel ECG data stream quickly and accurately, or alternatively, such a solution can be conveniently employed for real-time monitoring on a lightweight portable device. The model is tested on two open-access ECG databases: The China Physiological Signal Challenge (2020) database (CPSC-DB) with more than one million beats, and the commonly used MIT-BIH Arrhythmia Database (MIT-DB). Experimental results demonstrate that the proposed systematic approach achieves 99.30% F1-score, 99.69% recall, and 98.91% precision in CPSC-DB, which is the best R-peak detection performance ever achieved. Results also demonstrate similar or better performance than most competing algorithms on MIT-DB with 99.83% F1-score, 99.85% recall, and 99.82% precision. Compared to all competing methods, the proposed approach can reduce the false-positives and false-negatives in Holter ECG signals by more than 54% and 82%, respectively. Finally, the simple and invariant nature of the parameters leads to a highly generic system and therefore applicable to any ECG dataset.

ПОШУКОВЕ ДОСЛІДЖЕННЯ ПРЕДИКТОРІВ ПІЗНІХ РЕЦИДИВІВ АРИТМІЇ У ПАЦІЄНТІВ З ПЕРСИСТУЮЧОЮ ФІБРИЛЯЦІЄЮ ПЕРЕДСЕРДЬ ПІСЛЯ КАРДІОВЕРСІЇ

Introduction. The existing guidelines do not contain a clear algorithm for predicting the late recurrences of atrial fibrillation (AFib). Objectives. We have studied the predictors of late recurrence of arrhythmia in patients with persistent AFib after the restoration of sinus rhythm (SR). Research methods. A prospective single-center study included 120 hospitalized patients with persistent AFib who underwent successful cardioversion. The recurrence of AFib was assessed in the early period after cardioversion by recording 12-channel ECG, Holter ECG monitoring and event ECG monitoring. After a nine-month period follow-up visit was performed. Baseline demographic, clinical-functional features and comorbidities were compared in group with (n = 87) and without (n = 33) late recurrences of Afib. Results. There were no differences in demographic, anthropometric data and comorbidities. Patients with late recurrence of AFib had a lower heart rate (HR) after SR recovery (p &lt;0.001). Early recurrences of AFib were found in 43 (49.4%) patients with late recurrences of arrhythmia and only in two (6.0%) without late recurrences (p &lt;0.001). Patients with late recurrences of arrhythmia had a lower average daily HR (p &lt;0.001), a higher number of single atrial extrasystoles (p = 0.001), atrial pairs and runs (p = 0.01) recorded by Holter ECG monitoring. Conclusions. Late recurrences of arrhythmia during the nine-month follow-up period were reported in 72.5% of patients. Groups of patients with and without late recurrence of arrhythmia had significant differences in the incidence of early recurrences, mean HR on SR, and the presence of atrial extrasystoles after rhythm recovery.

Automatic Detection System of ECG Performance Based on Single Channel ECG Recorder

Aiming at the ECG performance indicators in the YY 0885-2013 "Medical electrical equipment Part 2: Particular requirements for the safety including essential performance of ambulatory electrocardiographic systems", the traditional detection methods are time-consuming and the test results are greatly affected by human factors. In this paper, an automated detection method is proposed. A set of automatic detection software for ECG performance is designed and developed based on a single-channel ECG recorder, and an automated detection system is set up in combination with standard testing equipment. And then, the MSA tool is used to analyze the repeatability and stability of the detection system, and the results show that the detection system is acceptable, and it can improve detection efficiency.

Abstract 14086: A Hybrid Machine Learning Method for Instantaneous Classification of Left Ventricular Filling Pressure Using Femoral Waveforms

Introduction: Non-invasive and instantaneous evaluation of left ventricular end-diastolic pressure (LVEDP) is of high significance in diagnosis and treatment of heart failure patients. We recently proposed the cardiac triangle mapping (CTM), a systems approach that uses information from arterial pressure waveforms and pre-ejection periods to map the global ventricular function (Pahlevan et al. Fluids 4.1 (2019): 16). Here, we propose a hybrid machine learning (ML) method based on CTM theory that uses pressure waveforms at the iliac bifurcation level and ECG to instantaneously classify the level of LVEDP. Methods: We studied 46 patients (Age: 39-90 (66.4±9.9), BMI: 20.2-36.8 (27.6±4.1), 12 females) who were scheduled for clinical left heart catheterization or coronary angiogram at the Keck Medical Center of USC. Exclusion criteria were valvular heart disease, left bundle branch block, or atrial fibrillation. Invasive LVEDP and pressure waveforms at iliac bifurcation were measured using a 3F Millar transducer tipped catheter, with simultaneous 3 channel ECG. The range of patients’ LVEDP was from 9.3 to 40.5 mmHg. LVEDP=18 mmHg was used as the ML classification cut-off. A random forest classifier (of n=30 grown decision trees, ensemble learning method: bootstrap aggregation, 6 inputs) was trained using data from 36 patients. The model was blindly tested on 10 additional patients. Results: Our ML model showed accuracies of 100.0% and 80.0% for predicting the true class of LVEDP on blind test data (i.e., elevated or normal, respectively) (Fig1). The accuracy on all data was higher than 93.4% for all classes. Conclusions: We demonstrated the proof-of-concept that a physics-based ML model can instantaneously classify the LV filling pressure using information extracted from femoral waveform and ECG. Our study is invasive validation; however, all required ML input parameters can be obtained noninvasively and remotely (i.e., using wearable devices or a smartphone).

Determination of left ventricular diastolic dysfunction using machine learning methods

Abstract Introduction Nowadays, more than 50% of new cases of heart failure (HF) are due to heart failure with preserved ejection fraction (HFpEF). Diastolic dysfunction of left ventricular underlies the development of HFpEF. Determination of diastolic function can be determinated using echocardiographic examination but not as a method of screening. Purpose Create a Machine Learning model to determinate severe (2nd or 3rd degrees) left ventricular diastolic dysfunction (LV DD) using single channel ECG. Methods The study prospectively included 375 patients aged between 18 and 88 years. Each of them underwent an assessment of left ventricular diastolic function using echocardiography based on ASE/EACVI guidelines and standard 2016. ECG was registered in I standard lead for 3 minutes. Then a spectral analysis of the ECG was performed using a wavelet transform based on the Fourier transform. Prediction of LV DD was implemented based on different machine learning methods. Results Based on the time intervals between the waves, the energy of the ECG signal in the target zones of the complexes, the amplitude at different points of the ECG complexes and the asymmetry indicators, different mathematical models were created to determine the severe (2nd or 3rd degree) LV DD. Lasso regression predicted the presence of severe LV DD with sensitivity 90% and specificity 80,6% (AUC- 0.856). The most significant for this method were the following variables: TA, SDNN, RonsF, ASEPMAX, B1, TpTe, ADP, SEPMAX, Sbeta, SEPF, RR, SRP-B0, Pan.1, QT/TQ, PpeakN, SA, SRP-B1. Random Forest also showed high sensitivity 80% and specificity 89, 3% (AUC- 0.860). For Random Forest model the most dignificant were: TpTe, VAT, QTc, QT/TQ, J80A, TA, Tenergy, Tpenergy, Beta, QRS11energy, Pan, Pan.1, RA, Pfi, QRSst, QRSfi, PpeakP, PpeakN, Rpeak, Speak, RonsF, B0-B1, SEP, SEP-B1, SEP-B0, SRP, SRP-B1, SRP-B0, ASEP. LV DD 1 degree did not have markers in our study. Conclusion Several machine learning methods have shown high sensitivity and specificity in identifying severe LV DD. Potentially ECG can be used as a screening method for determination of the LV DD. Funding Acknowledgement Type of funding sources: Public Institution(s). Main funding source(s): This work was financed by the Ministry of Science and Higher Education of the Russian Federation within the framework of state support for the creation and development of World-Class Research Centers “Digital biodesign and personalized healthcare” No. 075-15-2020-926