Measurements of pulse rate using facial video cameras from smart devices in patients diagnosed with atrial fibrillation.

Clinical applications of passive long-term heart rate (HR) monitoring in patients with cardiac arrhythmias include adequate drug titration of atrioventricular (AV) nodal drugs and assessment of medical compliance with treatment. A majority of patients treated with beta-blockers, especially patients with atrial fibrillation (AF), require some degree of drug titration during the first 6 months of treatment to ensure that adequate HR control and medicine compliance has been achieved. Failing to achieve adequate rate control in patients with AF can lead to worsening symptoms, heart failure exacerbations, and potentially tachycardia-induced cardiomyopathy. Enabling video-based monitoring during telehealth patient visits could facilitate providers to measure heart rate (HR) without the need for a dedicated home device (smartwatch, SPO2 device, or others). Videoplethysmography (VPG) is a monitoring technology that measures pulse rate by utilizing front-facing cameras embedded in smart devices. VPG provides a remote and contactless cardiac monitoring solution. We conducted a clinical experiment to evaluate the accuracy of VPG in measuring HR while running on two portable devices: Samsung S10 smartphones and S3 tablets. We used a single­lead ECG to measure the heart rate at the time of the VPG recordings in AF patients. We employed the Bland-Altman method to measure the level of agreement between videoplethysmography and ECG-based measurements of HR. The findings reveal that the mean difference in videoplethysmography and ECG-based heart rate was inferior to 1 bpm across the 2 devices with confidence intervals ranging from 3 to 12 BPM. Our facial video-based HR monitoring solution could assist providers in measuring heart rates in their patients with AF during remote telehealth visits.

The multiple arrhythmia dataset evaluation database (M.A.D.A.E.).

The convergence between wearable and medical device technologies is a natural progression. Miniaturization has allowed the design of small, compact monitoring systems that can record physiological signals over longer periods of time. Thus, the potential for these devices to expand the understanding of disease progression and patients' clinical status is very high. The accuracy of these devices, however, is dependent upon the computer algorithms utilized in the analysis of the large volume of physiological data monitored and/or recorded by the devices. Automated interpretation of the data by these new technologies, therefore, necessitates closer examination by regulatory organizations. The current requirements for the validation of novel Ambulatory ECG (A-ECG) annotation algorithms are based on the AAMI/ANSI-EC57 and IEC60601-2-47 Standard. These standards are being updated, but they rely on a very limited set of digitized ECG recordings from a couple of ECG databases built in the first half of the 70's. These reference signals are obsolete. We are developing a validation tool for computerized methods designed to detect and monitor cardiac activities based on body-surface ECGs. We will rely on a set of existing digital high-resolution 12­lead A-ECG recordings acquired in cardiac patients and healthy individuals. These ECG signals include a large and unique set of electrocardiographic events. This tool is being qualified by the Center for Devices and Radiological Health of the United States Food and Drug Administration (FDA) as a Medical Device Development Tool (MDDT). This document provides insights into the design of the M.A.D.A.E. database, its functionalities, and its ultimate role in enabling the next generations of automatic interpretation of ECG signals.

Semi-automated QRS score as a predictor of survival in CRT treated patients with strict left bundle branch block.

BACKGROUND: Cardiac Resynchronization Therapy (CRT) is widely used for treating selected heart failure patients, but patients with myocardial scar respond worse to treatment. The Selvester QRS scoring system estimates myocardial scar burden using 12-lead ECG. This study's objective was to investigate the scores correlation to mortality in a CRT population. METHODS AND RESULTS: Data on consecutive CRT patients was collected. 401 patients with LBBB and available ECG data were included in the study. QuAReSS software was used to perform Selvester scoring. Mean Selvester score was 6.4, corresponding to 19% scar burden. The endpoint was death or heart transplant; outcome was analyzed using Cox proportional hazards models. A Selvester score >8 was significantly associated with higher risk of the combined endpoint (HR 1.59, p=.014, CI 1.09-2.3). CONCLUSION: Higher Selvester scores correlate to mortality in CRT patients with strict LBBB and might be of value in prognosticating survival.

T-wave morphology can distinguish healthy controls from LQTS patients.

Long QT syndrome (LQTS) is an inherited disorder associated with prolongation of the QT/QTc interval on the surface electrocardiogram (ECG) and a markedly increased risk of sudden cardiac death due to cardiac arrhythmias. Up to 25% of genotype-positive LQTS patients have QT/QTc intervals in the normal range. These patients are, however, still at increased risk of life-threatening events compared to their genotype-negative siblings. Previous studies have shown that analysis of T-wave morphology may enhance discrimination between control and LQTS patients. In this study we tested the hypothesis that automated analysis of T-wave morphology from Holter ECG recordings could distinguish between control and LQTS patients with QTc values in the range 400-450 ms. Holter ECGs were obtained from the Telemetric and Holter ECG Warehouse (THEW) database. Frequency binned averaged ECG waveforms were obtained and extracted T-waves were fitted with a combination of 3 sigmoid functions (upslope, downslope and switch) or two 9th order polynomial functions (upslope and downslope). Neural network classifiers, based on parameters obtained from the sigmoid or polynomial fits to the 1 Hz and 1.3 Hz ECG waveforms, were able to achieve up to 92% discrimination between control and LQTS patients and 88% discrimination between LQTS1 and LQTS2 patients. When we analysed a subgroup of subjects with normal QT intervals (400-450 ms, 67 controls and 61 LQTS), T-wave morphology based parameters enabled 90% discrimination between control and LQTS patients, compared to only 71% when the groups were classified based on QTc alone. In summary, our Holter ECG analysis algorithms demonstrate the feasibility of using automated analysis of T-wave morphology to distinguish LQTS patients, even those with normal QTc, from healthy controls.

The use of beat-to-beat electrocardiogram analysis to distinguish QT/QTc interval changes caused by moxifloxacin from those caused by vardenafil.

QT correction factors (QTc) can cause errors in the interpretation of drug effects on cardiac repolarization because they do not adequately differentiate changes when heart rate or autonomic state deviates from the baseline QT/RR interval relationship. The purpose of our study was to determine whether the new method of QT interval dynamic beat-to-beat (QTbtb) analysis could better discriminate between impaired repolarization caused by moxifloxacin and normal autonomic changes induced by subtle reflex tachycardia after vardenafil. Moxifloxacin produced maximum mean increases of 13-14 ms in QTbtb, QTcF, and QTcI after 4 h. After vardenafil administration, a 10-ms effect could be excluded at all time points with QTbtb but not with QTcF or QTcI. Subset analysis of the vardenafil upper pharmacokinetic quartile showed that the upper bound of QTcF and QTcI was >10 ms, whereas that of QTbtb was <8 ms. This study demonstrated that newer methods of electrocardiogram (ECG) analysis can differentiate changes in the QT interval to improve identification of proarrhythmia risk.

Analysis of T-wave Amplitude Adaptation to Heart Rate Using RR-binning of Long-Term ECG Recordings.

The prognosis of patients with coronary artery disease at the early stage of the disease is a challenge of modern cardiology. There is an urgent need to risk stratify these patients. Holter technology is a cheap and cost effective tool to evaluate electrical abnormalities in the heart. We propose to investigate T-amplitude adaptation to heart rate (HR) using RR-binning. We used daytime recordings from healthy subjects and subjects with acute myocardial infarction (AMI) from the Telemetric and Holter ECG Warehouse. The AMI subjects were divided into two groups based on location of their infarction (group A: anterior or anterior lateral, group B: inferior or inferior lateral). Both AMI groups had acute and stable phase recordings. Population-based T-adaptation to HR was observed for healthy subjects (R2 = 0.92) but was less pronounced for AMI subjects: [Formula: see text].

A Comparison of IIR and Wavelet Filtering for Noise Reduction of the ECG.

This study compares the ability to preserve information and reduce noise contaminants on the ECG for five wavelet filters and three IIR filters. Two 3-lead Holter ECGs were used. White Gaussian Noise was added to the first ECG in increments of 10% coverage. The second ECG contained alternating muscle transients and noise-free segments. Computation times and SNR improvements for different noise coverages were calculated and compared. RMS errors were calculated from noise-free segments on the ECG with transient muscle noise. Wavelet filters improved SNR more than IIR filters when the signal coverage was more than 50% noise. In contrast, the computation times were shorter for IIR filters (6 s) than for wavelet filters (88 s). On the ECG with transient muscle noise there was a trade-off in performance between wavelet and IIR filtering. In a clinical setting where the amount of noise is unknown, using IIR filters appears to be preferred for consistent performance.

A Wavelet-Based Algorithm for Delineation and Classification of Wave Patterns in Continuous Holter ECG Recordings.

Quantitative analysis of the electrocardiogram (ECG) requires delineation and classification of the individual ECG wave patterns. We propose a wavelet-based waveform classifier that uses the fiducial points identified by a delineation algorithm. For validation of the algorithm, manually annotated ECG records from the QT database (Physionet) were used. ECG waveform classification accuracies were: 85.6% (P-wave), 89.7% (QRS complex), 92.8% (T-wave) and 76.9% (U-wave). The proposed classification method shows that it is possible to classify waveforms based on the points obtained during delineation. This approach can be used to automatically classify wave patterns in long-term ECG recordings such as 24-hour Holter recordings.

Electrocardiographic method for identifying drug-induced repolarization abnormalities associated with a reduction of the rapidly activating delayed rectifier potassium current.

Several important non-cardiac drugs have been removed from the market after revealing harmful effect that was not identified during prior safety-assessment studies. We developed a new technique for the measurements of repolarization abnormalities from surface ECGs; this method improves sensitivity and specificity of the current technique used to identify the presence of abnormal ion current kinetics in the myocardial cells namely a prolongation of the QT interval on the surface ECG signal. We described in this paper the method and preliminary results, revealing the superiority of our technique that may play a role in the future of drug-safety assessment.

Heart rate variability in patients with congenital long QT syndrome.

BACKGROUND: The congenital long QT syndrome (LQTS) affecting myocardial repolarization is caused by mutations in different cardiac potassium or sodium channel genes. Adrenergic triggers are known to initiate life-threatening torsade de pointes ventricular tachycardias in LQTS patients, and anti-adrenergic therapy has been shown to be effective in many cases. Despite this well-documented adrenergic component, the data about autonomic modulation of the heart rate in LQTS, as described by heart rate variability (HRV) analysis, are very limited. METHODS: Conventional time- and frequency-domain and newer nonlinear measures of HRV were compared in resting conditions among 27 LQTS patients with gene mutations at the LQT1 (n = 8), LQT2 (n = 10) or LQT3 (n = 9) loci and 34 LQTS noncarrier family members. RESULTS: None of the conventional time- or frequency-domain or newer nonlinear measures of HRV differed significantly between the LQTS carriers and LQTS noncarriers or between the LQT1, LQT2, and LQT3 carriers. CONCLUSIONS: These findings suggest that baseline cardiac autonomic modulation of the heart rate measured in resting conditions by traditional or newer nonlinear measures of HRV is not altered in LQTS patients. Furthermore, no differences are observed in HRV parameters between LQTS patients with potassium (KvLQT1, HERG), and sodium (SCN5A) ion channel gene mutations. HRV analysis in resting conditions does not improve phenotypic characterization of LQTS patients.

Cardiovascular effects of air pollution: what to measure in ECG?

Epidemiologic evidence indicates that air pollution adversely affects the cardiovascular system, leading to increased cardiovascular morbidity and mortality. However, the mechanisms of such an association are unknown. Although potential mechanisms of deleterious effects of air pollution may involve response of the respiratory system, immunologic response, or coagulation abnormalities, the cardiovascular system seems to be the common end point of these pathways. Cardiovascular response to any stress (which may include air pollution) is a consequence of a complex interplay between the autonomic nervous system governing centrally mediated control of the cardiovascular system, a myocardial substrate (current state of the myocardium) altered in the course of disease processes, and myocardial vulnerability leading to arrhythmogenic or ischemic response. Through the use of standard electrocardiograms (ECGs), exercise ECG testing, and long-term ambulatory ECG monitoring, modern electrocardiology makes a valuable contribution to understanding the different mechanistic factors involved in the increase in adverse cardiovascular events due to air pollution. Heart rate variability analysis can provide quantitative insight into the autonomic response of the cardiovascular system to air pollution. Analysis of ventricular repolarization in an ECG (both duration and morphology) gives valuable information about the status and dynamic behavior of myocardium, reflecting myocardial substrate and vulnerability. ST-segment analysis of ECGs is used routinely to monitor the magnitude of ischemia and could be used to monitor subtle changes in the myocardium in subjects exposed to air pollution. Comprehensive analysis of ECG parameters describing the influence of the autonomic nervous system, the role of myocardial substrate, and the contribution of myocardial vulnerability could and should be employed in air pollution studies, especially as those mechanistic components have been proven to contribute to increased cardiovascular morbidity and mortality in general.

Fractal correlation properties of heart rate dynamics and adverse events in patients with implantable cardioverter-defibrillators.

The aim of this study was to determine the prognostic significance of nonlinear and standard heart rate (HR) variability parameters in predicting future adverse events (AEs) in patients with implantable cardioverter-defibrillators. In postinfarction studies, nonlinear measures of HR variability obtained from long-term electrocardiographic recordings have been suggested to be better predictors of adverse outcomes than conventional HR variability measures. Fifty-five high-risk patients with reduced left ventricular function and an implantable cardioverter-defibrillator had a 10-minute, high-resolution electrocardiographic recording after which they were followed for 25 months on average. Implantable cardioverter-defibrillator shock or death was determined as the end point. The SD of all normal-to-normal RR intervals, the square root of the mean squared differences of successive normal-to-normal RR intervals, and the proportion of interval differences of successive normal-to-normal RR intervals >50 ms, low-frequency and high-frequency powers of the power spectrum and their ratio were calculated as conventional measures of HR variability. The short-term scaling exponent (alpha(1)) and approximate entropy were determined as nonlinear measures of HR variability. AEs occurred in 23 patients (42%). Patients with AEs had significantly lower alpha(1) than event-free patients: 0.81 +/- 0.29 (mean +/- SD) versus 1.01 +/- 0.30 (p = 0.02). None of the other HR variability parameters differed significantly between patients with and without AEs. In the Cox proportional-hazards model including age, gender, ejection fraction, occurrence of ventricular tachyarrhythmia before defibrillator implantation, beta-blocker usage, and alpha(1), only alpha(1) was an independent predictor of AEs: hazard ratio 1.20 (95% confidence interval 1.03 to 1.39) for every 0.10 decrease in alpha(1) (p = 0.020). In conclusion, alpha(1) obtained from a 10-minute electrocardiographic recording yields important prognostic information about the risk of AEs in patients with implantable cardioverter-defibrillators.

Identification of post-myocardial infarction patients prone to ventricular tachycardia using time-frequency analysis of QRS and ST segments.

BACKGROUND: Late potentials (LPs) in the terminal portion of the QRS complex are commonly sought to identify post-myocardial infarction patients prone to ventricular tachyarrthythmias (VT) or sudden death. More recent time frequency signal processing tools have been shown to provide new parameters for the quantification of LPs and abnormal activities buried within the QRS complex. METHODS AND RESULTS: The study population comprised 23 myocardial infarction patients with documented sustained VT (MI+VT), 40 myocardial infarction patients without VT (MI - VT) and 31 normal subjects. The reproducibility of the method was tested in an additional set of 66 patients. The signal-averaged high-resolution electrocardiograms (HRECGs) were quantified by deconstructing the unfiltered X, Y and Z leads using a 511-orthogonal wavelet network. Using receiver operating characteristics (ROC) curves and discriminant analysis applied to the wavelet coefficients, we extracted the most significant wavelets to classify the post MI patients. These wavelets detected time-frequency alterations both in the ST segment and within the QRS complex, characterizing patients prone to VTs. The same statistical methods were applied to the conventional time-domain measurements. The combined application in our population of the orthogonal wavelet deconstruction method and discriminant analysis had 91% sensitivity and 95% specificity, an improvement of 22% and 25%, respectively, compared with the conventional time domain method. Reproducibility was 82%. CONCLUSIONS: In post-myocardial infarction patients, orthogonal wavelet transforms can detect alterations in high-frequency components within the QRS and ST segment. Our findings support the view that wavelet-related parameters are more relevant than those of the time domain method in predicting subsequent malignant tachyarrhythmias.

Beat-to-Beat repolarization variability in LQTS patients with the SCN5A sodium channel gene mutation.

Current techniques evaluating beat-to-beat variability of repolarization rely on accurate determination of T wave endpoints. This study proposes a T wave endpoint-independent method to quantify repolarization variability in a standard 12-lead ECG using a wavelet transformation. Our method was used to identify repolarization variability in long QT syndrome patients (LQTS) with the SCN5A sodium channel gene mutation. Using wavelet transformations based on the second Gaussian derivative, we evaluated repolarization variability in 11 LQTS patients with the mutation, 13 noncarrier family members, and 28 unrelated healthy subjects. Time-domain repolarization variability parameters (SDRTo, SDRTm) and wavelet parameters describing temporal (beat-to-beat) variability of repolarization in time (TVT) and in amplitude (TVA) were analyzed. Reproducibility of wavelet parameters and relationship of wavelet-based variability with heart rate and preceding RR interval were investigated. The wavelet-based method quantified beat-to-beat variability of the entire repolarization segment (regardless of QT interval identification) providing insight into variability in repolarization morphology. Our method showed that SCN5A carriers have significantly increased repolarization variability in amplitude (23% +/14% vs 8 +/- 4%, P < 0.001) and in time (14 +/- 17 ms vs 3 +/-2 ms, P < 0.004) compared to noncarriers. Variability of repolarization amplitude was found to be heart rate dependent with variability decreasing with increasing heart rate. Relative error describing reproducibility of TVA and TVT was < or = 5% and < or =10%, respectively. Our method quantifies repolarization variability in amplitude and in time without the need to identify T or U wave endpoints. Wavelet-detected repolarization variability contributes to phenotypic identification of SCN5A carriers, with more pronounced beat-to-beat variability in repolarization amplitude than in time.

ECG features of microvolt T-wave alternans in coronary artery disease and long QT syndrome patients.

UNLABELLED: T-wave alternans (TWA) is a marker of myocardial electrical instability. We compared ECG features of microvolt TWA in coronary artery disease (CAD) and long QT syndrome (LQTS) patients. METHOD: The study populations consisted of 43 CAD and 39 LQTS patients. TWA was detected in resting Holter recordings using the new correlation method (CM). After preprocessing to adjust for RR variability and respiratory modulation, CM was used to quantify TWA amplitude (A(CM)), duration (N(CM)), and magnitude (MAG(CM); defined as the product of A(CM) and N(CM)). RESULTS: TWA was detected in 19 (44%) CAD and 17 (44%) LQTS patients. TWA was associated with longer RR intervals (P = 0.006) and had larger magnitudes (P = 0.067) in LQTS than CAD patients. The TWA was identified as transient (nonstationary) in 15 of 19 (79%) TWA-positive CAD patients, and in 8 of 17 (47%) TWA-positive LQTS patients (P = 0.047). CONCLUSIONS: The frequency of TWA detected with CM is similar in LQTS and CAD patients. TWA is larger in LQTS than in CAD patients, whereas TWA is more frequently transient (nonstationary) in LAD than LQTS patients. In LQTS patients, but not in CAD patients, a longer RR is associated with TWA, indicating different electrophysiologic mechanisms in the two pathologies.

Are serial Holter QT, late potential, and wavelet measurement clinically useful?

Clinical centers are increasingly using new techniques such as Holter QT, late potential, and wavelet measurements. However, we lack validated databases for the assessment of the performance of the signal-processing methods and their reproducibility. Failure of the QT interval to adapt to changes in the heart rate is considered to be a more meaningful parameter than QT prolongation itself. In this study, different factors that may affect the reproducibility of QT and QTm (onset of the QRS to the maximum of T) measurement are analyzed: the incidence of sympathetic tone and parasympathetic activity on low- and high-frequency QT variability, the very low frequency dependency of the QT interval to changes in the R-R interval, changes in the heart's position, and measurement errors. Typical root-mean-square values of the beat-to-beat measurement errors in upright-position Holter recordings are only 1.5 ms for QT versus 3.4 ms for QTm. Although the dependence of the QT interval on the heart rate is well established, the method for rate correction of the QT interval remains controversial. None of the formulas for heart rate adjustment of the QT previously proposed provide complete correction for all of the rate influences involved due to "memory phenomenon"; that is, there is a time delay, ranging up to 3-4 minutes, between a change in heart rate and the subsequent change in the QT interval. This problem has been solved by developing patient-specific neural networks that are trained to "identify" the dynamic behavior of the QT interval (or QTm) as a function of the R-R interval in order to predict the beat-to-beat changes of the QT interval as a function of the measured beat-to-beat changes of the R-R interval. Computing the differences between the predicted and the measured QT interval will allow for the detection of any significant deviations, both in the steady-state and transient conditions. Recent developments in the analysis of the high-resolution electrocardiogram (HRECG) in the time domain and frequency domain, with emphasis on the assessment of the reproducibility of late potential and wavelet measurements, are also reported in this study. The two main causes of variability in HRECG analysis are physiology and, for time-domain analysis, intermanufacturer variability. Physiologic changes can be overcome by standardizing the clinical protocols and repeating the recordings. The most important technical requirement for the proper use of late potentials is to standardize the algorithm for the detection of QRS offset among different late potential analyzing machines so that clinical data can be exchanged. The recently introduced wavelet transform provides a fruitful alternative to the more classical time-domain methods. Preliminary results show an 8 to 15% performance improvement over conventional time-domain analysis for the stratification of the HRECG after myocardial infarction. Reproducibility is excellent, up to 100%, but needs to be assessed on larger populations matched for age, sex, and pathology.

Stratification of time-frequency abnormalities in the signal-averaged high-resolution ECG in postinfarction patients with and without ventricular tachycardia and congenital long QT syndrome.

Having developed sound mathematical techniques that allow precise mapping of cardiac signals in the time-frequency (TF) and time-scale planes, the next important issue is to extract from these representations information that best reflects the electrophysiologic and anatomic derangement unique to patients at risk of arrhythmias and other cardiac diseases. In this study, the authors present a new method that stratifies the magnitude of the TF transforms of abnormal cardiac signals into distinguishing features by comparing the means of the coefficients of the TF transforms of any study population to the corresponding means of a control population using a standard ANOVA technique. This results in a three-dimensional mapping of the high-resolution ECG into time, frequency, and P value components. Significant energy increases are given positive P values and depressed energies are given negative P values: these are ranked according to a color scale. The method was tested on two study populations: postmyocardial infarction patients with documented ventricular tachycardia (MI+VT, n = 23) and without (MI-VT, n = 40) and patients with congenital long QT syndrome (LQTS, n = 19). Two groups of healthy control subjects (n = 31 and n = 40) were used as a reference group matched for sex. The study results were based on the Morlet analyzing wavelets, with frequencies ranging from 40 to 250 Hz in 10 logarithmically progressing scales, and computed millisecond per millisecond over a 350-ms analyzing time window, starting from 100 ms before the onset of the QRS. The patients with MI+VT displayed significantly increased high-frequency components in the 40-250-Hz frequency range, corresponding to prolonged QRS duration and late potentials in the area from 80 to 150 ms after QRS onset. Significantly depressed energy (P < 10(-4)) was also observed for the 40-106-Hz frequency range in the first 50 ms of the QRS complex, mainly in lead Y and in the magnitude vector. In patients with LQTS, significant modifications (P < 10(-2)) were observed in the first half of the QRS and in the ST-segment, in all leads, revealing anomalies in the genesis of the ventricular depolarization and repolarization processes. In conclusion, the authors propose a new method for the stratification of abnormal TF components occurring in the signal-averaged high-resolution electrocardiogram of patients at risk of VT and fibrillation under different pathologic conditions.