Short ECG Research Articles

K-B2S+: A one-dimensional CNN model for AF detection with short single-lead ECG waves from wearable devices

Wearable signal analysis is an important technology for monitoring physiological signals without interfering with an individual's daily behavior. As detecting cardiovascular diseases can dramatically reduce mortality, arrhythmia recognition using ECG signals has attracted much attention. In this paper, we propose a single-channel convolutional neural network to detect Atrial Fibrillation (AF) based on ECG signals collected by wearable devices. It contains 3 primary modules. All recordings are firstly uniformly sized, normalized, and Butterworth low-pass filtered for noise removal. Then the preprocessed ECG signals are fed into convolutional layers for feature extraction. In the classification module, the preprocessed signals are fed into convolutional layers containing large kernels for feature extraction, and the fully connected layer provides probabilities. During the training process, the output of the previous pooling layer is concatenated with the vectors of the convolutional layer as a new feature map to reduce feature loss. Numerous comparison and ablation experiments are performed on the 2017 PhysioNet/CinC Challenge dataset, demonstrating the superiority of the proposed method.

Recognition of life-threatening arrhythmias by ECG scalograms

This work is devoted to the automatic classification of six classes of life-threatening arrhythmias using short ECG fragments of 2s-length. This task is extremely important for the detection of life-threatening arrhythmias with continuous monitoring. Especially dangerous are ventricular fibrillation and high-frequency heartbeat ventricular tachycardia. Timely detection of these dangerous disorders in the clinic allows doctors to effectively use electrical defibrillation, which saves the patient's life. A feature of our approach is the use of a unique technique for converting ECG signals into images (scalograms) using a continuous wavelet transform. For arrhythmia classification, the AlexNet neural network with a well-known deep learning architecture, which is commonly used in image classification tasks, is used. The experiments used data from the PhysioNet database, as well as synthesized ECG data obtained using the SMOTE method. The experimental results show that the proposed approach allows achieving an average accuracy of 98.7% for all classes, which exceeds the maximum accuracy estimates of 93.18% previously obtained by other researchers.

Association between Val158Met polymorphism in the COMT gene with anxiety and heart rate variability in children with irritable bowel syndrome

The aim of this study was to examine the peculiarities of trait anxiety and heart rate variability parameters as well as their relationship depending on the catechol O-methyltransferase (COMT) gene Val158Met polymorphism in children with irritable bowel syndrome. Materials and methods. A total of 26 children aged 6–12 years with a verified diagnosis of irritable bowel syndrome according to the Rome IV criteria were examined. Beforehand, all the patients underwent molecular genetic testing for the COMT Val158met single nucleotide polymorphism by using the polymerase chain reaction-restriction fragment length polymorphism method. Heart rate variability was analyzed via cardiointervalography (Neurosoft) based on short five-minute resting-state ECG recordings. The CMAS (Children’s Manifest Anxiety Scale) test was used to measure trait anxiety levels. The Pearson’s test was used to assess correlations between heart rate variability parameters and anxiety levels. Data were processed using Microsoft Excel 2016 and analyzed with GraphPad (Prism 5.0). Results. Depending on the functional COMT Val158met polymorphism, all the patients were allocated to 3 groups: 8 children with 472 GA (Val/Met) COMT genotype; 10 children with 472 AA (Met/Met) genotype; 8 children with 472 GG (Val/Val) genotype. Time and frequency domain parameters of heart rate variability were significantly different in each group. The highest level of anxiety and the largest percentage of LF component (mainly sympathetic activity) in the structure of heart rate variability was noted among Met/Met carriers. Val/Val carriers had a significantly lower anxiety level and an autonomic imbalance with a higher percentage of HF component (parasympathetic activity). Positive correlations between trait anxiety and heart rate variability parameters were found only in Val/Val and Val/Met groups. Conclusions. Our study has revealed the influence of the COMT Val158met polymorphism on the level of trait anxiety and heart rate variability parameters. It is important for a better understanding of the gut-brain axis dysregulation and impaired stress resilience in children with irritable bowel syndrome. Also, these data could be used to improve current schemes for the treatment of irritable bowel syndrome, supplementing them with activation therapy, psychotherapy, psychopharmacotherapy.

AN EXPLORATORY STUDY OF PHYSIOLOGICAL AND BIOCHEMICAL PARAMETERS TO IDENTIFY SIMPLE, ROBUST AND RELEVANT BIOMARKERS FOR THERAPEUTIC INTERVENTIONS FOR PTSD: STUDY RATIONALE, KEY ELEMENTS OF DESIGN AND A CONTEXT OF WAR IN UKRAINE

Introduction. The incidence of post-traumatic stress disorder (PTSD) is increasing due to war in Ukraine is of growing importance as this disorder is a serious risk factor for alcohol/drug abuse, depression, suicide and functional dysregulation of internal organs, in particular cardiovascular, endocrine, immune systems. PTSD can also lead to social inadaptation in family and at work. This makes PTSD a very important social problem in Ukraine that causes significant economic damage, since most of the PTSD patients are young and in working age.
 Objectives. The purpose of this paper is to describe the rationale, key elements of design, methodology and future prospects of the ongoing project “Exploratory study of health related physiological and biochemical parameters to identify simple, robust and relevant biomarkers for therapeutic interventions for PTSD patients”.
 Methods. Short time ECG records (5 minutes in supine position and 6 minutes in orthostatic test) recorded by “CARDIOLAB” (Kharkiv, Ukraine) digital ECG device with software to analyze HRV was used to calculate time-domain and frequency-domain parameters of HRV.
 Results. The pattern of HRV changes indicates suppression of parasympathetic activity, low overall spectral power and dominance of sympathetic branch of autonomic nervous system in participants with PTSD.
 Conclusions. PTSD has no effective treatment so far and the best clinical practices rarely reach efficacy of 50%. The realization of the exploratory clinical study described in this paper will provide insights in the mechanisms underlying systemic consequences of PTSD and will build the evidence enabling clinical trial(s) for studying treatment intervention for patients with PTSD.

Automatic diagnosis of cardiovascular diseases using wavelet feature extraction and convolutional capsule network

Cardiovascular diseases (CVDs) are the primary cause of high mortality levels worldwide with coronary artery disease (CAD) being the largest contributor. Delay in diagnosis of CAD may progress and lead to myocardial infarction and congestive heart failure. Thus, timely diagnosis is a crucial key to prevent the development of serious complications and allow an optimum medication. This paper intends to design a diagnosis system to classify-eight different diseases using a capsule network with wavelet decomposed images of short ECG segments. It is noteworthy that in contrast to the previous studies that require large training data to efficiently predict unseen data, our designed models achieve high performance with less training data and training time. In the proposed system, firstlyECG signals are preprocessed and segmented into beats segments. Then, the continuous wavelet transform is used to transform the segments into2D scalograms. These latter are decomposed using discrete wavelet transform to obtain wavelet coefficient imageswhich are fed to the reformed capsule networks. The focal loss was applied to mitigate the class-imbalance issue. The proposed model was trained and tested using 5-fold cross-validation and train-test split techniques. The best performance parameters were achieved in the train-test split approach with an accuracy of 98.6%, sensitivity of 99.6%, specificity of 98.7%, precision of 99.1%, F1 score of 99.4% and area under the roc value of 0.99.Furthermore, experimental results demonstrated the superiority of our approach over state-of-the-art techniques using the same dataset.

Abstract 154: Limitation of Troponin T In Identifying the Cause of Out-of-hospital Cardiac Arrest

Introduction: High Sense-Troponin T (HS-TnT) is a key determinant for presence or absence of acute MI in chest pain patients (pts). Its value for acute MI diagnosis in pts with out-of-hospital cardiac arrest (OHCA) needs assessment because myocardial ischemia may be the cause but also a consequence of cardiac arrest causing global ischemia during no and low flow periods. Methods: Data were prospectively collected. From pts included in the ARREST registry who survived OHCA to discharge with VF as initial rhythm, the cause of OHCA and peak HS-TnT were analyzed. Cause of OHCA was determined independently by researchers, from history, admission and serial ECG, coronary angiography (CAG), cardiac imaging and genetics as available and indicated, without using HS-TnT information. MI diagnosis was made according to the 4 th universal definition of MI: type 1 MI (thrombotic STEMI/NSTEMI), type 2 MI (supply/demand imbalance) and no MI (short transient ECG changes and all non-ischemic causes). Results: From 2012 till 2018 in the study region 373 OHCA pts met the inclusion criteria and had a known peak HS-TnT. Of these, Type 1 MI was present in 109 STEMI and 39 NSTEMI pts, type 2 MI in 116 pts and no MI in 109 pts. No significant coronary stenosis or normal CAG was present in 81 pts: 11/106 (10%) with type 2 MI and 70/109 (64%) without MI. HS-TnT exceeded the conventional diagnostic threshold of 4 times upper limit of normal (4x14ng/L = 56 ng/L) in 99%, 95%, 95% and 89% of pts, respectively. Median peak HS-TnT levels in STEMI, NSTEMI, Type 2 and no AMI pts was 1890 ng/L, 480 ng/L, 300 ng/L and 261 ng/L, respectively. (Figure) Area Under the ROC-Curve of HS-TnT values to differentiate between type 1 NSTEMI and type 2 AMI was 0.649, between type 1 NSTEMI and no MI 0.684 and between type 2 MI and no AMI 0.555. Conclusion: Exclusion of diagnosis of MI in OHCA pts cannot be made with HS-TnT, nor a distinction between NSTEMI, type 2 MI and no AMI. Based on HS-TnT many OHCA pts without ischemic cause would classify as NSTEMI.

Machine Learning Methods in Predicting Patients with Suspected Myocardial Infarction Based on Short-Time HRV Data.

Diagnosis of cardiovascular diseases is an urgent task because they are the main cause of death for 32% of the world’s population. Particularly relevant are automated diagnostics using machine learning methods in the digitalization of healthcare and introduction of personalized medicine in healthcare institutions, including at the individual level when designing smart houses. Therefore, this study aims to analyze short 10-s electrocardiogram measurements taken from 12 leads. In addition, the task is to classify patients with suspected myocardial infarction using machine learning methods. We have developed four models based on the k-nearest neighbor classifier, radial basis function, decision tree, and random forest to do this. An analysis of time parameters showed that the most significant parameters for diagnosing myocardial infraction are SDNN, BPM, and IBI. An experimental investigation was conducted on the data of the open PTB-XL dataset for patients with suspected myocardial infarction. The results showed that, according to the parameters of the short ECG, it is possible to classify patients with a suspected myocardial infraction as sick and healthy with high accuracy. The optimized Random Forest model showed the best performance with an accuracy of 99.63%, and a root mean absolute error is less than 0.004. The proposed novel approach can be used for patients who do not have other indicators of heart attacks.

Assessment of the pre-test probability of ischemic heart disease according to the data of dispersion mapping of an electrocardiogram and analysis of heart rate variability

February 2022 est., there were several validated ischemic heart disease (IND) pretest probability scales presented in the clinical guidelines of various cardiology communities (G.A. Diamond and J.S. Forrester, Duke, ESC 2019 scales). Despite their diversity, they lacked clear quantitative criteria and were based on a subjective assessment of the pain syndrome described by a patient, their gender and age. The purpose of this study was to investigate the possibilities of heart rate variability and ECG dispersion mapping for pre-test assessment of the probability of coronary artery disease. We studied 81 patients (mean age 61.48 ± 13.00 years) with suspected CAD. All patients underwent a five-minute ECG recording with the construction of ECG dispersion mapping and assessment of heart rate variability. Each subject underwent coronary angiography (within 12 months before or after the stress test) to verify the damage to the coronary bed. When analyzing the comparison of the results of coronary angiography and dispersion ECG mapping, 3 groups of patients were formed and studied: group A (coronary artery stenosis more than 50%, n = 18), B (coronary artery stenosis less than 50%, n = 16) and C (without coronary artery disease, n = 21). It was noted that the more pronounced the stenosis of the coronary arteries, the higher the index of microalternations “Myocardium” and T-alternation of the myocardium, and the lower the functional reserve of the myocardium. Considering that a short (thirty second) high-resolution ECG recording is required to perform dispersion ECG mapping, this method can be used as a screening for the selection of patients at high risk of coronary heart disease and exercise testing. We suggest that patients should be referred for exercise testing if two of the three criteria are present: BMI ≥ 22%, Functional reserve ≤ 70%, T-alternation ≥ 18% (AUC ROC is 0.718). The sensitivity and specificity of the presented method can be increased by adding indicators from the analysis of heart rate variability (heart rate, HF, LF) and patient’s age (AUC ROC is 0.929) to the formula.

Exploring novel algorithms for atrial fibrillation detection by driving graduate level education in medical machine learning

During the lockdown of universities and the COVID-Pandemic most students were restricted to their homes. Novel and instigating teaching methods were required to improve the learning experience and so recent implementations of the annual PhysioNet/Computing in Cardiology (CinC) Challenges posed as a reference. For over 20 years, the challenges have proven repeatedly to be of immense educational value, besides leading to technological advances for specific problems. In this paper, we report results from the class ‘Artificial Intelligence in Medicine Challenge’, which was implemented as an online project seminar at Technical University Darmstadt, Germany, and which was heavily inspired by the PhysioNet/CinC Challenge 2017 ‘AF Classification from a Short Single Lead ECG Recording’. Atrial fibrillation is a common cardiac disease and often remains undetected. Therefore, we selected the two most promising models of the course and give an insight into the Transformer-based DualNet architecture as well as into the CNN-LSTM-based model and finally a detailed analysis for both. In particular, we show the model performance results of our internal scoring process for all submitted models and the near state-of-the-art model performance for the two named models on the official 2017 challenge test set. Several teams were able to achieve F1 scores above/close to 90% on a hidden test-set of Holter recordings. We highlight themes commonly observed among participants, and report the results from the self-assessed student evaluation. Finally, the self-assessment of the students reported a notable increase in machine learning knowledge.

The effects of lead combination on short and long vector ECG

The effects of lead combination on short and long vector ECG

QT interval extracted from 30-minute short resting Holter ECG recordings predicts mortality in heart failure

BackgroundProlonged repolarization duration is a significant total mortality (TM) predictor in post-myocardial infarction patients. AimWe examined the clinical significance of QT interval that was extracted from a Short Resting Holter ECG (SRH ECG – 30-min duration) as a TM predictor in heart failure (HF) patients. MethodsOne hundred forty-five HF patients (male = 84%, mean age = 64 ± 12 years, mean LVEF = 33 ± 10%) underwent an SRH ECG recording for 30 min. These high-resolution ECG signals were analyzed and the QT interval was calculated and corrected according to the Fridericia formula. After 42.1 months, 26 patients died. ResultsUnivariate analysis for Deceased and Living groups: QTc:455 ± 33 ms vs 441 ± 32 ms (p = 0.04), LVEF:32 ± 10% vs 34 ± 9% (p < 0.5), Mean Heart Rate: 73 ± 11 bpm vs 69 ± 12 bpm (p = 0.2), SDNN/HRV: 45 ± 42 ms vs 41 ± 29 ms (p = 0.4), QRS: 123 ± 26 ms vs 119 ± 29 ms (p = 0.5). Multivariate Cox regression analysis with model adjusted for QTc, Mean Heart Rate, LVEF, QRS, revealed that QTc-Fridericia interval was an independent TM predictor (H.R.:1.017, 95% C.I.: 1.003–1.030, p = 0.01). The cut-off point of 490 ms (90th percentile) in the same model presented HR: 2.9 for TM (95%C.I.: 1.066–7.882, p = 0.03). Kaplan Meier curves depicted a clear difference in survival between the two patients' groups (QTc Group≥490 ms vs QTc Group <490 ms). The curve diverge was important (log-rank, p = 0.02). ConclusionA fast risk stratification approach with SRH ECG recording is an efficient method for flash evaluation of mortality risk in HF patients.

24 h Holter ECG Monitoring of Patients with Rheumatoid Arthritis-A Potential Role for a Precise Evaluation of QT Interval Duration and Associated Arrhythmic Complications.

Background: Patients with rheumatoid arthritis (RA) have increased systemic inflammatory burden associated with elevated cardiovascular mortality. Prolonged ventricular repolarisation evaluated by QT interval duration is a risk factor for cardiovascular and total mortality. In RA, mortality risk is correlated with dynamics and cumulative incidence of QTc prolongation rather than QTc value. The aim is to evaluate if QT parameters evaluated with 24 h Holter ECG are a better option to complete the cardiovascular profile of RA patients than parameters from short ECG recordings. Materials and methods: A total of 58 patients (22 males, 36 females) with RA were submitted to short ECG recordings at admission and to 24 h Holter ECG. QT interval parameters and ventricular ectopy generated from both types of recordings were analyzed. Results: QTc interval values obtained from Holter ECG were significantly higher than the values from short term ECG and were correlated with severity of inflammatory process. The number of QRS complexes with QTc > 450 ms recorded during 24 h Holter was strongly correlated with the number of ventricular events and severity of the inflammatory process. Conclusions: In patients with RA, the Holter ECG recordings could realize a more precise evaluation of the extent and dynamics of QTc interval duration and of ventricular ectopic events with potential risk of sudden death.

Correction to: Inter-patient detection of atrial fibrillation in short ECG segments based on LSTM network with multiple input layers

Correction to: Inter-patient detection of atrial fibrillation in short ECG segments based on LSTM network with multiple input layers

Inter-patient detection of atrial fibrillation in short ECG segments based on LSTM network with multiple input layers

Inter-patient detection of atrial fibrillation in short ECG segments based on LSTM network with multiple input layers

Air Pollution Role as Risk Factor of Cardioinhibitory Carotid Hypersensitivity

Little is known about the impact of air pollution on neuroautonomic system. The authors have investigated possible influence of air pollution and outdoor temperature on the carotid sinus hypersensitivity (CSH), as main cause of neurally mediated syncope in forty-years-old subjects and older. Pollutants’ concentrations and outdoor temperature of days in which 179 subjects with recurrent syncope underwent carotid sinus massage (CSM) were analyzed. Before this manoeuvre, cardiovascular control by short period heart and blood pressure spectral duration of segment between the end of P and R ECG-waves (PeR) were registred; RR variability on the same short period ECG recordings and their spectral coherence were also analyzed. CSH was found in 57 patients (28 with cardioinhibitory response and 29 subjects showed vasodepressor reaction), while 122 subjects had a normal response. CSM performed during high ozone concentrations was associated with slightly higher risk of cardioinhibitory response (odd ratio 1.012, 95% CI 1.001–1.023, p &lt; 0.05), but neither this or other polluting agent nor outdoor temperature seemed to influence autonomic control in basal resting condition. Thus, ozone seemed to influence response to the CSM in CSH patients and it is probably able to facilitate a cardioinhibitory response, perhaps through an increase of nerve acetylcholine release. P→PR coherence could be useful in predicting a sinus cardioinhibitory hypersensitivity in those cases when CSM is contraindicated.

Hybrid feature fusion for classification optimization of short ECG segment in IoT based intelligent healthcare system

Hybrid feature fusion for classification optimization of short ECG segment in IoT based intelligent healthcare system

Multiclass Classification of Cardiac Rhythms on Short Single Lead ECG Recordings using Bidirectional Long Short-Term Memory Networks

The recognition of cardiac rhythms is a topic of great relevance, particularly using short single-lead ECG recordings, due to its potential to early detect cardiovascular diseases and take actions quickly to preserve peoples wellbeing. Among cardiac arrhythmias, atrial fibrillation is the most common sustained cardiac arrhythmia, with significant mortality and morbidity rates. Several approaches have been conceived to identify cardiac rhythms, from the comparison of heart rate with adaptive and fixed thresholds to the application of deep and machine learning techniques. In this work, the classification of three cardiac rhythms (normal, atrial fibrillation, and other arrhythmias), as well as the identification of noise recordings, were performed using bidirectional LSTM networks that exploit the ECG signal (a representation of the cardiac electric activity) and time series containing information about the auricular and ventricular activities. A Monte Carlo 10-fold cross-validation of 10 iterations was performed to assure the generalization of the classifiers and the replicability of the results. An average accuracy of 77.97% was obtained to recognize the four classes but increase up to 85.95% when noise recordings were left out of the classification process. Moreover, micro F1 scores of 89.96%, 79.23%, and 79.77% were obtained for normal rhythm, atrial fibrillation, and other arrhythmias, respectively. The imbalance of classes and the characteristic patterns of normal rhythm and atrial fibrillation were the main factors associated with these performances.

Effect of year-round training on parameters of heart rate variability, central hemodynamics and physical working capacity in short-distance swimmers

Dynamic examination of athletes is of great interest for sports medicine specialists, as it allows measuring changes in athletes’ functional state in parallel with increasing athletic performance in a one-year training cycle. The aim of the work is to assess dynamic changes in parameters of heart rate variability, central hemodynamics and physical performance in short-distance swimmers during preparatory and competitive periods of the training process. Materials and methods. The swimmers (n = 94) qualified from the second-class sport qualification to Master of Sports of International Class were examined during the preparatory and competitive periods of the training process. Short ECG recordings of 5 minutes were used for the heart rate variability (HRV) analysis according to the International Standard. Results. In the preparatory period, the autonomic equilibrium index had a mean value of 3.104 ± 0.494 1/s2, and in the competitive period – 2.459 ± 0.248 1/s2 (P = 0.295). The initial distribution of athletes depending on the value of autonomic tone, according to the classification proposed by R. M. Bayevskiy, had the following ratio in the preparatory period: 75.0 % / 20.0 % / 5.0 %, vagotonic, normotonic and sympathotonic individuals, respectively, statistically significantly indicating the prevalence of athletes with vagotonia over normotonic individuals (P = 0.0005). In the competitive period, this ratio was 60.0 % / 40.0 % / 0 %. The central hemodynamic indices showed a downward trend as evidenced by a decrease in the cardiac index from 2.954 ± 0.126·l·min-1·m-2 to 2.862 ± 0.088 l‧min-1‧m-2 (Р = 0.510), although these values matched to the eukinetic circulatory type (CT) during both training periods. However, the ratio of hypokinetic, eukinetic and hyperkinetic CT in the preparatory period was 45.0 % / 45.0 % / 10 0 %, and in competitive period – 45.0 % / 55.0 % / 0 %, respectively. The mean value of physical working capacity (PWC170/kg) in the preparatory period was 15.54 ± 0.66 kgm·min-1·kg-1, and in the competitive period it was 18.09 ± 0.53 kgm·min-1·kg-1, making an increase of 16.41 % (Р = 0.0001). The functional state index (FSI) increased significantly by 25.08 % from 5.558 ± 0.322 r. u. to 6.952 ± 0.272 r. u. (Р = 0.00001). The correlation analysis during the preparatory period showed a significant positive correlation between Mo and PWC170/kg (r = 0.46, P = 0.040), ARI and CI (r = 0.50, P = 0.026), IARP and CI (r = 0.53, P = 0.017), stress-index and SI (r = 0.52, P = 0.019) and negative – between Mo and CI (r = -0.56, P = 0.009). In the competitive period, the general tendency of correlation interaction was remained, as evidenced by revealed positive relationships between Mo and FSI (r = 0.53, P = 0.017), ARI and SI (r = 0.51, P = 0.020). Conclusions. Short-distance swimmers in the competitive period had significantly higher level of physical working capacity (by 16.4 %), the functional state index (by 25.1 %), as well as parasympathetic tone of ANS in comparison with the preparatory period. The above-mentioned hemodynamic changes and autonomic balance indicators were associated with the increase in athletic performance: 4 (20 %) swimmers qualified for the title of Master of Sports of Ukraine, 2 (10 %) swimmers – for Candidate Master of Sports, and 2 (10 %) swimmers – for first-class sport qualification.

Automatic atrial fibrillation detection from short ECG signals: A hybrid deep learning approach

Atrial fibrillation (AF) is one of the most common arrhythmic complications. Recently, researchers have attempted to use deep learning models, such as convolution neural networks (CNN) and/or Long Short-Term Memory (LSTM) neural networks to alleviate the tedious and time-consuming feature extraction process and achieve good classification results. In this paper we propose a hybrid CNN-LSTM model and use the short ECG signal from the PhysioNet/CinC Challenges 2017 dataset to explore and evaluate the relative performance of four data mining algorithms and three deep learning architectures. The original ECG signal, clinical diagnostic features and 169 features based on time domain, frequency domain and non-linear heart rate variability indicators were used for comparative experiments. The results show that with proper design and tuning, the Hybrid CNN-LSTM model performed much better than other benchmarked algorithms. It achieves 97.42% accuracy, 95.65% sensitivity, 97.14% specificity, 0.99 AUC (Area under the ROC curve) value and 0.98 F1 score. In general, with proper design and configuration, deep learning can be effective for automatic AF detection while data mining methods require domain knowledge and an extensive feature extraction and selection process to get satisfactory results. However, most machine learning algorithms, including deep learning models, perform the task as a black box, making it almost impossible to determine what features in the signal are critical to the analysis.

A FEATURE SELECTION-BASED ALGORITHM FOR DETECTION OF ATRIAL FIBRILLATION USING SHORT-TERM ECG

In the presence of premature atrial contraction (PAC), premature ventricular contraction (PVC) or other ectopic beats, RR intervals (RRIs) may be disturbed, which results in other types of heart disease being misdiagnosed as atrial fibrillation (AF). In this study, a low-complexity AF detection method based on short ECG is proposed, which includes RRIs modification and feature selection. The extracted RRIs are used to determine whether the potential RRI interference exists and to modify it. Next, based on the modified RRIs, the features are evaluated and selected by the methods of correlation criterion, Fisher criterion, and minimum redundancy maximum relevance criterion. Finally, filtered features are classified by the artificial neural network (ANN). The algorithm is validated in a test set including 2332 AF, 313 normal (NOR), 239 atrioventricular block (IAVB), 81 left bundle branch block (LBBB), 624 right bundle branch block (RBBB), 426 PAC and 564 PVC. Compared with the previous detection method of AF based on the RRIs, the proposed method achieved an overall sensitivity of 94.04% and an overall specificity of 86.74%. The specificity of the test set containing only AF and NOR is up to 99.04%. Meanwhile, the overall false-positive rate (FPR) of PAC and PVC can be reduced by 9.19%. While ensuring accuracy, this method effectively reduces the probability of misdiagnosis of PVC and PAC as AF. It is an automatic detection method of AF suitable for inter-patient clinical short-term ECG.