Digital 12-lead Electrocardiograms Research Articles

Smaller intrinsic QRS area and smaller reduction of QRS area during cardiac resynchronisation therapy are associated with higher risk of ventricular arrhythmias and appropriate ICD therapy

Abstract Background Cardiac resynchronisation therapy (CRT) patients are at high risk for ventricular arrhythmias (VAs), and little is known about which characteristics are associated with an increased risk for arrhythmic events. Purpose To investigate the association between QRS area parameters and the risk for VAs in patients with heart failure undergoing primary preventive CRT with defibrillator (CRT-D) implantation. Methods All patients receiving CRT-D between 2015 and 2020 at a large-volume tertiary care center were retrospectively evaluated. Digital 12-lead electrocardiograms (ECGs) were collected before and after CRT implantation. Patients were included if they had native LBBB and were implanted with a primary preventive CRT-D. Kors’ regression transformation was used to derive vectorcardiographic QRS area from the 12-lead ECGs. QRS area parameters were analysed in relation to a primary endpoint of first appropriate ICD therapy, i.e., ATP or shock therapy. Results 177 patients with LBBB (68.9 years [60.7–73.8]; 21.5% female; 52.9% New York Heart Association (NYHA) class III-IV; LVEF 25.9±6.3%) were followed over a median follow-up time of 2.5 years [1.6–3.0] after CRT-D implantation. 27 patients reached the primary endpoint. The median pre-CRT QRS area was 134.5 μVs [102.3–163.8]. Multivariable Cox regression identified a small pre-CRT QRS area (HR, per 10 μVs decrease, 1.30; [1.12–1.51]; p=<.001) and a small post-CRT reduction in QRS area (HR, per 10 μVs decrease, 1.30; [1.13–1.49]; p=<.001) as independent predictors of an increased risk of appropriate ICD therapy. In analyses with a scoring variable combining pre-CRT QRS area and QRS area reduction both above or below median, the combination of a small pre-CRT QRS area and a small QRS area reduction was strongly associated with an increased risk of reaching the endpoint (HR, per point increase, 2.47; [1.2–4.9]; p=0.01). Conclusion A small intrinsic QRS area, by itself or in combination with a small reduction of post-CRT QRS area, was strongly associated with a higher risk of ventricular arrhythmias in this cohort of CRT-D treated patients. In conjunction with other parameters, further risk stratification for arrhythmic events could be achieved by assessing QRS area data pre- and post-CRT.Figure 1.K-M pre-CRT QRS area quartilesFigure 2.K-M QRS area score

The Role of P Wave Parameters in Predicting Pulmonary Vein Isolation Outcomes for Paroxysmal Atrial Fibrillation: An Observational Cohort Study.

Pulmonary vein isolation (PVI) is an effective management method for paroxysmal atrial fibrillation (PAF). The P wave in the 12-lead electrocardiogram (ECG) represents atrial depolarisation. This study aims to utilise the P wave to predict PVI outcomes for PAF. This single-centre retrospective study aimed to predict PVI outcomes using P wave parameters. It included 211 consecutive patients with first PVI for PAF between 2018 and 2019 and targeted the pulmonary veins (PVs). Procedure success was defined by freedom of ECG-documented AF at 12 months. Digital 12-lead ECGs with 1-50 hertz bandpass filters were monitored before the procedure. Corrected P wave duration (PWDc), P wave amplitude (PWV), P wave dispersion (PWDisp), intra-atrial block (IAB), P wave area (PWA), and P wave terminal force in V1 (PTFV1) were measured before ablation and correlated with the outcomes. Successful PVI occurred in 154 patients (73%). Demographics were similar between both arms. P wave parameters correlated with PVI failure included increased PWDc in all leads except for lead III, aVR, and V3, decreased PWV in lead I (hazard ratio [HR]: 0.7, 95% confidence interval [CI]: 0.53-0.95), lead II (HR: 0.45, 95% CI: 0.22-0.65), aVL (HR: 0.58, 95% CI: 0.22-0.98), and aVF (HR: 0.67, 95% CI: 0.58-0.87), decreased PWA in lead I (HR: 0.55, 95% CI: 0.21-0.76), lead II (HR: 0.48, 95% CI: 0.34-0.87), aVL (HR: 0.65, 95% CI: 0.45-0.96), and aVF (HR: 0.61, 95% CI: 0.32-0.89), and the presence of IAB (HR: 2, 95% CI: 1.4-4.2, p = 0.02). PWDisp and PTFV1 were not correlated with PVI outcome. PWDc, PWA, PWV, and IAB are valuable predictors for PVI outcome for PAF at 12 months.

Comparison of electrocardiographic parameters between left bundle optimized cardiac resynchronization therapy (LOT-CRT) and left bundle branch pacing-cardiac resynchronization therapy (LBBP-CRT).

In patients undergoing cardiac resynchronization therapy using left bundle branch area pacing (LBBP-CRT), the addition of a coronary sinus lead, that is, Left bundle optimized CRT (LOT-CRT) might confer additional benefits. To compare the electrocardiographic characteristics between LBBP-CRT and LOT-CRT MATERIALS AND METHODS: Patients with non-ischemic cardiomyopathy (NICMP) and left bundle branch block (LBBB) with left ventricular ejection fraction<35% who underwent implantation of an atrial lead, a left bundle lead, and a coronary sinus lead were included in this prospective study. Digital 12-lead electrocardiograms were recorded in three pacing modes-AAI, DDD with pacing from the LBB lead (LBBP-CRT), and DDD with pacing from both left bundle and coronary sinus leads (LOT-CRT). QRS duration (QRSd), QRS area, QT interval, and T peak-T end (TpTe) intervals were compared. Among 24 patients, QRSd reduced from 167±21.2ms to 134.5±23.6ms with LBBP-CRT (p<.001) and 129.5±18.6ms with LOT-CRT (p<.001) without a significant difference between LBBP-CRT and LOT-CRT (p=.15). Patients with QRS duration with LBBP-CRT>131ms showed a significant reduction in QRSd with LOT-CRT (p=.03). QT interval was reduced with both modes of CRT. LOT-CRT was associated with a greater reduction in QRS area (p=.001), TpTe interval (p=.03), and TpTe/QT ratio (p=.013) compared to LBBP-CRT. In patients with NICMP and LBBB, there was no significant difference in QRSd with LOT-CRT compared to LBBP-CRT. However, in patients with QRSd>131ms after LBBP-CRT, LOT-CRT resulted in a significantly narrower QRS.

Adavosertib (AZD1775) does not prolong the QTc interval in patients with advanced solid tumors: a phase I open-label study

PurposeAdavosertib is a small-molecule, ATP-competitive inhibitor of Wee1 kinase. Molecularly targeted oncology agents have the potential to increase the risk of cardiovascular events, including prolongation of QT interval and associated cardiac arrhythmias. This study investigated the effect of adavosertib on the QTc interval in patients with advanced solid tumors.MethodsEligible patients were ≥ 18 years of age with advanced solid tumors for which no standard therapy existed. Patients received adavosertib 225 mg twice daily on days 1–2 at 12-h intervals and once on day 3. Patients underwent digital 12-lead electrocardiogram and pharmacokinetic assessments pre-administration and time-matched assessments during the drug administration period. The relationship between maximum plasma drug concentration (Cmax) and baseline-adjusted corrected QT interval by Fridericia (QTcF) was estimated using a prespecified linear mixed-effects model.ResultsTwenty-one patients received adavosertib. Concentration–QT modeling of ΔQTcF and the upper limit of the 90% confidence interval corresponding to the geometric mean of Cmax observed on days 1 and 3 were below the threshold for regulatory concern (not > 10 ms). No significant relationship between ΔQTcF (vs baseline) and adavosertib concentration was identified (P = 0.27). Pharmacokinetics and the adverse event (AE) profile were consistent with previous studies at this dose. Eleven (52.4%) patients experienced 17 treatment-related AEs in total, including diarrhea and nausea (both reported in six [28.6%] patients), vomiting (reported in two [9.5%] patients), anemia, decreased appetite, and constipation (all reported in one [4.8%] patient).ConclusionAdavosertib does not have a clinically important effect on QTc prolongation.ClinicalTrials.govNCT03333824.

Normal values of electrocardiogram among healthy young adult African population

Abstract Background: The study aimed at establishing the normal value of electrocardiographic parameters in healthy young African population, in sub-Saharan Africa, using digital electrocardiogram (ECG) recordings and comparing between males and females. Methods: The study population included 656 healthy young adults from the ages of 15–40 years, undergraduate students, made up of 357 females and 299 males. The study took place in the Department of Human Physiology Laboratory of our University. The 12-lead ECGs were recorded and read using a validated computer program. Various parameters were recorded and the 2nd and 98th percentiles were taken as the normal range of the study population. Results: Findings from our study show some differences from studies made in the same region using manual methods of ECG readings and compared to studies done in other races. The 2nd percentile, median, and 98th percentile of the studied parameters were as follows: PR interval was 102 ms, 140 ms, and 196 ms, respectively, in females and 104 ms, 144 ms, and 202 ms, respectively, in males. P-wave duration was 76 ms, 100 ms, and 126 ms in females and 62 ms, 98 ms, and 130 ms in males, respectively. We found the QTc (Bazett) to be 371 ms, 420 ms, and 467 ms in females and 344 ms, 389 ms, and 442 ms in males, respectively. Frontal QRS axis was − 6°, 53°, and 113° in females and − 11°, 57°, and 108° in males; frontal T axis was − 8°, 36°, and 141° in females and − 127°, 48°, and 75° in males; and QRS-T angle was 8°, 43°, and 115° in females and 25°, 84°, and 153° in males, respectively. Conclusion: This study provided reference values for healthy young Sub-Saharan African population using digital 12-lead ECG. It will serve as a reference for what is normal in our race and also prognosticate and monitor those that are predisposed to developing some cardiovascular diseases in the future.

Abstract 9422: Tandem Deep Learning and Logistic Regression Models to Optimize Hypertrophic Cardiomyopathy Detection in Routine Clinical Practice

Background: An electrocardiogram (ECG)-based artificial intelligence (AI) algorithm has shown good performance in detecting hypertrophic cardiomyopathy (HCM AI-score). However, its clinical application may be challenging due to the low disease prevalence and potentially high false-positive rate when screening unenriched cohorts. Aims: To identify clinical characteristics associated with false-positive HCM AI-score results to improve screening tool application. Methods: From a cohort of 20,677 patients with a digital 12-lead ECG in routine clinical practice in 2021, we performed a detailed chart review of the 200 patients with highest HCM AI-score to determine their HCM status and collect clinical data. Stepwise forward logistic regression was used to create a clinical variable-based “candidacy for HCM screening” score (CHS score), differentiating true- and likely true-positive from likely false-positive HCM AI-score results. We then validated the CHS score in 200 patients with the highest HCM AI-scores from an independent cohort of 15,147 patients with an ECG in 2022. Results: Among the 200 patients in 2021 (median age 71 years, 48% female), 96 (48%) were likely false-positive HCM AI-score results, 71 (36%) definite or likely true-positive results, and 33 (16%) indeterminate. The CHS score included a history of surgical septal myectomy, coronary artery bypass graft surgery, cardiac valve replacement, and pacemaker, and had an AUC of 0.88 (95% CI 0.83-0.92) for the differentiation of true- versus false-positive AI results. In the 2022 cohort, the CHS score had an AUC of 0.82 (95% CI 0.76 - 0.88). After limiting the cohort to HCM candidates identified by the CHS score, the false-positive results rate was reduced to 27/114 (24%). Conclusions: By applying the CHS score to patients with positive HCM AI-screening results, the false-positive rate decreased from 49% to 24%. Utilizing a clinical model used in tandem with an AI model improved diagnostic yield.

Comparative diagnostic and prognostic implications of right versus left bundle branch block in patients with suspected ST-segment elevation myocardial infarction

Abstract Background Current ESC guidelines on management of acute myocardial infarction in patients presenting with ST-segment elevation (STEMI) recommend emergent invasive management in patients with ischemic symptoms and bundle branch block (BBB). While left bundle branch block (LBBB) is a well-known high-risk feature in patients with acute myocardial infarction, data on the diagnostic and prognostic impact of right bundle branch block (RBBB) in comparison to LBBB in the context of STEMI are scarce. Methods A prospective cohort of 2,139 patients with suspected STEMI were triaged by the on-call cardiologist to acute coronary angiography based on a prehospital, tele-transmitted digital 12-lead electrocardiogram. All discharge diagnoses were independently adjudicated. The diagnostic accuracy of RBBB or LBBB was calculated. 30-day and 1-year all-cause mortality were from a central national registry and compared for LBBB, RBBB and patients without BBB using the Kaplan Meier method. Unadjusted and adjusted (age, sex, hypertension, hyperlipoproteinemia, smoking status and history of coronary artery disease) hazard ratios for 1-year mortality were calculated for both BBB-types. Results The final diagnosis of STEMI was adjudicated in 1,832/2,139 (85.6%) of patients. RBBB was present in 117 (5.5%), LBBB in 61 (2.9%) patients, respectively. Patients with BBB were older than non-BBB patients: Median age was 69.7 years in RBBB, 68.9 years in LBBB and 62.6 years in non-BBB patients. Comparing both BBB types, LBBB patients more often had prior heart failure (6.3% in RBBB vs 11.5% in LBBB), while RBBB patients more often developed cardiac arrest before angiography (15.3% vs 3.7% in RBBB and LBBB patients respectively) and more frequently revealed Thrombolysis in Myocardial Infarction (TIMI) grade flow of 0 (61.0% vs 47.2%, respectively). Specificity for the final diagnosis of STEMI was not significantly different in both BBB-types: 95.1% (95% CI 92.1–97.0) in RBBB versus 92.8% (95% CI 89.4–95.2) in LBBB patients (p=0.25). Patients with RBBB had significantly higher 30-day and 1-year mortality (Figure 1), while LBBB and non-BBB patients showed similar outcome. Unadjusted hazard ratio for 1-year mortality comparing RBBB to non-BBB patients was 3.35 (95% CI 2.11–5.34, p&amp;lt;0.001) and 1.4 (95% CI 0.57–3.44, p=0.46) when comparing LBBB to non-BBB patients. After adjustment hazard ratio for RBBB patients was 2.3 (95% CI 1.25–4.21, p=0.007). Conclusion In the context of suspected STEMI, RBBB is an ominous sign associated with adverse clinical features like cardiac arrest prior to PCI, TIMI grade 0 flow, compared to LBBB or non-BBB presenters. Consequently, RBBB patients had poorer outcome after 12 months. Our data therefore reinforce current guidelines in designating RBBB-patients as very high-risk, demanding immediate management and triage. Funding Acknowledgement Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Regional Research Foundation in Region Sjaelland, Denmark

Screening for cardiac allograft rejection among heart transplant recipients using an electrocardiogram-based deep learning model

Abstract Background Current approaches utilizing non-invasive methods to screen for cardiac allograft rejection (gene expression profiling and cell free DNA) have yet to be broadly integrated into heart transplant management and have shown limited discrimination (AUCs of 0.68 and 0.77, respectively). Changes in the electrocardiogram (ECG) have been reported at the time of severe cardiac rejection, including low voltages and conduction abnormalities. It remains unknown if subtle ECG changes correlating with cardiac allograft rejection can be detected earlier using machine learning methods. Purpose We sought to develop an artificial intelligence (AI) model to detect cardiac allograft rejection based on the 12 lead ECG. Methods We identified all patients who underwent a heart transplant at 3 hospital sites within a single health system from Jan 1998 through Apr 2021 and extracted digital 12-lead ECG data as well as endomyocardial biopsy pathology results from the electronic medical record. We partitioned our data into a training (80%), validation (10%), and test set (10%) based on a group stratification sampling. Each patient was present in only one set and each set had a positivity rate of 2.6% with 6,074/758/758 ECGs belonging to 1,146/140/141 unique patients in each set respectively. Cardiac allograft rejection was defined as moderate or severe acute cellular rejection based on International Society for Heart and Lung Transplantation (ISHLT) guidelines. A convolutional neural network, using the 12-lead ECG data as input, was trained with hyperparameter optimization for regularization, learning rate adjustments, and class weights. Model performance metrics were based on the test data and estimated using the final model architecture. Results 1,587 heart transplant recipients who had at least one endomyocardial biopsy were evaluated for inclusion. We limited our sample to ECGs performed within 30 days of the biopsy date (7,590 ECGs, representing 1,425 unique patients). Our study population had a median age of 55.8 years and 28.7% were female. The median number of ECG-biopsy pairs per patient was 5. The majority of endomyocardial biopsy results were classified as none or mild rejection (97.1%), and 2.9% had moderate/severe rejection. The ECG-based AI model detected cardiac allograft rejection with an area under the receiver operative curve (AUC) of 0.84 in the test set. The sensitivity, specificity, positive and negative predictive values were 95%, 52.6%. 5.2% and 99.7% respectively. Conclusions An AI-ECG model appears to outperform novel non-invasive laboratory tests (gene expression profiling and cell free DNA) for detecting cardiac allograft rejection and does not require a blood draw or the additional complexities surrounding sample processing. This model relies on a readily available and relatively inexpensive test, the ECG. In addition, AI predictions can be made available within a few minutes following ECG acquisition. Funding Acknowledgement Type of funding sources: Private hospital(s). Main funding source(s): Mayo Clinic

Machine learning-based phenotyping and risk assessment of hypertrophic cardiomyopathy - linking ECGs, morphology and genotypes

Abstract Funding Acknowledgements Type of funding sources: None. Background Integrating clinical data to distinguish hypertrophic cardiomyopathy (HCM) phenotypes is relevant in clinical practice. Machine learning (ML) can help - deep learning (DL) networks can automate detection and segmentation of 12-lead electrocardiograms (ECGs), whereas unsupervised learning can group patients to compare ECG, imaging and genetic characteristics. The aim is to automate ECG morphology analysis from all 12 ECG leads and multiple beats, and relate this to HCM genotypes and imaging phenotypes. Methods The single-center cohort included phenotype- and genotype-positive (G+) HCM patients (n = 104) and their phenotype-negative relatives (n = 50, 42% G+). All patients had a digital 12-lead ECG, echocardiography, and a magnetic resonance (CMR) study performed. The workflow is shown in Fig 1. A U-Net DL network was used for ECG delineation (P, QRS, T onsets/offsets) for all cardiac cycles. Three heartbeats were selected for each patient based on their morphology, with the aim of capturing beat-to-beat variability. An unsupervised representation learning algorithm was used to fuse ECG data and assess inter-patient similarities. Patients were clustered based on similarities of ECG biomarkers, and compared with regards to genotypes, family history of sudden cardiac death (SCD), history of ventricular arrhythmias/syncope/aborted SCD, implanted defibrillators (ICD), left ventricular (LV) obstruction, maximal wall thickness, late gadolinium enhancement (LGE), and HCM risk-SCD score. Results ML based on ECG biomarkers provided a good separation of HCM patients and relatives (Fig 1A), also showing G- and patients with variants of uncertain significance grouping together (Fig 1B). Clustering resulted in 6 ECG phenogroups (C1-6). C1 and 2 were related to less comorbidities, cardiac remodeling, and HCM risk score, capturing the majority of G- patients. C3 and 4 were related to LV obstruction – where C4 consisted of symptomatic patients with high ICD implantation and event rates, high LGE, and impaired systolic function. C5 captured patients with high comorbidities, extremely remodeled hearts, but no obstruction, whereas C6 patients with positive family history and high arrhythmic events (Fig 1C, Table 1). The average ECG morphology is shown side-by-side for C1 and C5 in Fig 1D – negative T waves, increased R/S wave amplitudes, left axis deviation (LAD) and ST elevation can be recognized as macro-biomarkers in C5 (yellow arrows). Conclusion ML can automate the analysis of complex clinical data, simultaneously taking into account the morphology of all ECG components in all 12-leads, throughout multiple beats, compare it with clinical and imaging data, and identify clinically sensible phenogroups as validated by structural and functional findings, as well as with genotypes and clinical information. Automated and comprehensive cardiac data analysis has diagnostic and research potential to help screen populations and phenotype disease etiologies. Abstract Figure 1: analysis pipeline Abstract Table 1: clinical variables

Left Atrial Hypertension, Electrical Conduction Slowing, and Mechanical Dysfunction - The Pathophysiological Triad in Atrial Fibrillation-Associated Atrial Cardiomyopathy.

BackgroundAtrial fibrillation (AF) is the most common arrhythmia and a significant burden for healthcare systems worldwide. Presence of relevant atrial cardiomyopathy (ACM) is related to persistent AF and increased arrhythmia recurrence rates after pulmonary vein isolation (PVI).ObjectiveTo investigate the association of left atrial pressure (LAP), left atrial electrical [invasive atrial activation time (IAAT) and amplified p-wave duration (aPWD)] and mechanical [left atrial emptying fraction (LA-EF) and left atrial strain (LAS)] functional parameters with the extent of ACM and their impact on arrhythmia recurrence following PVI.Materials and MethodsFifty patients [age 67 (IQR: 61–75) years, 78% male] undergoing their first PVI for persistent AF were prospectively included. LAP (maximum amplitude of the v-wave), digital 12-lead electrocardiogram, echocardiography and high-density endocardial contact mapping were acquired in sinus rhythm prior to PVI. Arrhythmia recurrence was assessed using 72-hour Holter electrocardiogram at 6 and 12 months post PVI.ResultsRelevant ACM (defined as left atrial low-voltage extent ≥2 cm2 at <0.5 mV threshold) was diagnosed in 25/50 (50%) patients. Compared to patients without ACM, patients with ACM had higher LAP [17.6 (10.6–19.5) mmHg with ACM versus 11.3 (7.9–14.0) mmHg without ACM (p = 0.009)]. The corresponding values for the electrical parameters were 166 (149–181) ms versus 139 (131–143) ms for IAAT (p < 0.0001), 163 (154–176) ms versus 148 (136–152) ms for aPWD on surface-ECG (p < 0.0001) and for the mechanical parameters 27.0 (17.5–37.0) % versus 41.0 (35.0–45.0) % for LA-EF in standard 2D-echocardiography (p < 0.0001) and 15.2 (11.0–21.2) % versus 29.4 (24.9–36.6) % for LAS during reservoir phase (p < 0.0001). Furthermore, all parameters showed a linear correlation with ACM extent (p < 0.05 for all). Receiver-operator-curve-analysis demonstrated a LAP ≥12.4 mmHg [area under the curve (AUC): 0.717, sensitivity: 72%, and specificity: 60%], a prolonged IAAT ≥143 ms (AUC: 0.899, sensitivity: 84%, and specificity: 80%), a prolonged aPWD ≥153 ms (AUC: 0.860, sensitivity: 80%, and specificity: 79%), an impaired LA-EF ≤33% (AUC: 0.869, sensitivity: 84%, and specificity: 72%), and an impaired LAS during reservoir phase ≤23% (AUC: 0.884, sensitivity: 84%, and specificity: 84%) as predictors for relevant ACM. Arrhythmia recurrence within 12 months post PVI was significantly increased in patients with relevant ACM ≥2 cm2, electrical dysfunction with prolonged IAAT ≥143 ms and mechanical dysfunction with impaired LA-EF ≤33% (66 versus 20, 50 versus 23 and 55 versus 25%, all p < 0.05).ConclusionLeft atrial hypertension, electrical conduction slowing and mechanical dysfunction are associated with ACM. These findings improve the understanding of ACM pathophysiology and may be suitable for risk stratification for new-onset AF, arrhythmia recurrence following PVI, and development of novel therapeutic strategies to prevent AF and its associated complications.

Electrocardiographic Screening of Arrhythmogenic Cardiomyopathy in Genotype-Positive and Phenotype-Negative Relatives.

Background: Arrhythmogenic cardiomyopathy is a hereditary cause of ventricular arrhythmias and sudden death. Identifying the healthy genetic carriers who will develop the disease remains a challenge. A novel approach to the analysis of the digital electrocardiograms of mutation carriers through signal processing may identify early electrocardiographic abnormalities.Methods: A retrospective case–control study included a population of healthy genetics carriers and their wild-type relatives. Genotype-positive/phenotype-negative individuals bore mutations associated with the development of arrhythmogenic cardiomyopathy. The relatives included had a non-pathological 12-lead electrocardiogram, echocardiogram, and a cardiac magnetic resonance. Automatic digital electrocardiographic analyses comprised QRS and terminal activation delay duration, the number of QRS fragmentations, ST slope, and T-wave voltage.Results: Digital 12-lead electrocardiograms from 41 genotype-positive/ phenotype-negative (29 simple carriers and 12 double mutation carriers) and 73 wild-type relatives were analyzed. No differences in the QRS length, the number of QRS fragmentations, and the voltage of the T-wave were observed. After adjusting for potential confounders, double carriers showed an average ST-slope flatter than those of the simple carriers and wild type [5.18° (0.73–8.01), 7.15° (5.14–11.05), and 11.46° (3.94–17.49), respectively, p = 0.005]. There was a significant negative correlation between the ST slope and the age in genotype-positive/phenotype-negative relatives (r = 0.376, p = 0.021) not observed in their wild-type counterparts (r = 0.074, p = 0.570).Conclusions: A flattened ST segment may be an early sign of electrical remodeling that precedes T-wave inversion in healthy genetic carriers. A thorough analysis of the digital electrocardiographic signal may help identify and measure early electrical abnormalities.

The 12-lead electrocardiogram as a biomarker of biological age.

We have demonstrated that a neural network is able to predict a person's age from the electrocardiogram (ECG) [artificial intelligence (AI) ECG age]. However, some discrepancies were observed between ECG-derived and chronological ages. We assessed whether the difference between AI ECG and chronological age (Age-Gap) represents biological ageing and predicts long-term outcomes. We previously developed a convolutional neural network to predict chronological age from ECGs. In this study, we used the network to analyse standard digital 12-lead ECGs in a cohort of 25144 subjects ≥30 years who had primary care outpatient visits from 1997 to 2003. Subjects with coronary artery disease, stroke, and atrial fibrillation were excluded. We tested whether Age-Gap was correlated with total and cardiovascular mortality. Of 25144 subjects tested (54% females, 95% Caucasian) followed for 12.4 ± 5.3 years, the mean chronological age was 53.7 ± 11.6 years and ECG-derived age was 54.6 ± 11 years (R 2 = 0.79, P < 0.0001). The mean Age-Gap was small at 0.88 ± 7.4 years. Compared to those whose ECG-derived age was within 1 standard deviation (SD) of their chronological age, patients with Age-Gap ≥1 SD had higher all-cause and cardiovascular disease (CVD) mortality. Conversely, subjects whose Age-Gap was ≤1 SD had lower all-cause and CVD mortality. Results were unchanged after adjusting for CVD risk factors and other survival influencing factors. The difference between AI ECG and chronological age is an independent predictor of all-cause and cardiovascular mortality. Discrepancies between these possibly reflect disease independent biological ageing.

Automatic classification of healthy and disease conditions from images or digital standard 12-lead electrocardiograms

Standard 12-lead electrocardiography (ECG) is used as the primary clinical tool to diagnose changes in heart function. The value of automated 12-lead ECG diagnostic approaches lies in their ability to screen the general population and to provide a second opinion for doctors. Yet, the clinical utility of automated ECG interpretations remains limited. We introduce a two-way approach to an automated cardiac disease identification system using standard digital or image 12-lead ECG recordings. Two different network architectures, one trained using digital signals (CNN-dig) and one trained using images (CNN-ima), were generated. An open-source dataset of 41,830 classified standard ECG recordings from patients and volunteers was generated. CNN-ima was trained to identify atrial fibrillation (AF) using 12-lead ECG digital signals and images that were also transformed to mimic mobile device camera-acquired ECG plot snapshots. CNN-dig accurately (92.9–100%) identified every possible combination of the eight most-common cardiac conditions. Both CNN-dig and CNN-ima accurately (98%) detected AF from standard 12-lead ECG digital signals and images, respectively. Similar classification accuracy was achieved with images containing smartphone camera acquisition artifacts. Automated detection of cardiac conditions in standard digital or image 12-lead ECG signals is feasible and may improve current diagnostic methods.

Increases in repolarization heterogeneity predict left ventricular systolic dysfunction and response to cardiac resynchronization therapy in patients with left bundle branch block.

This study aimed to investigate the association between T-wave morphology and impaired left ventricular ejection fraction (LVEF) in patients with complete left bundle branch block (cLBBB), and the predictive value of T-wave morphology for response to cardiac resynchronization therapy (CRT). We enrolled 189 patients with cLBBB on electrocardiogram performed between January 2007 and December 2011 who underwent standard echocardiography. Repolarization parameters, including the QRS-to-T angle (TCRT), T-wave morphology dispersion (TMD), T-wave loop area (PL), and T-wave residuum (TWR), were reconstructed from digital standard 12-lead electrocardiograms by T-wave morphology analysis. CRT response was defined as ≥15% reduction in left ventricular end-systolic volume at 12 months after CRT implantation. The clinical outcome endpoint was a composite of heart failure hospitalization, heart transplantation, or death during follow up (mean, 5.8 years). On logistic regression, a higher heart rate, longer QRS duration, increased TMD, and larger TWR were all independently associated with LVEF < 40%. Among 40 patients who underwent CRT, those with a larger TMD (P = .007), larger PL (P = .025), and more negative TCRT (P = .015) had better response to CRT. A large TMD (P = .018) and large PL (P = .003) were also independent predictors of the clinical outcome endpoint. Increases in repolarization heterogeneity in patients with cLBBB are associated with impaired LVEF. A large TMD and large PL may be useful as additional predictors of response to CRT, improving patient selection for CRT.

Usefulness of Machine Learning-Based Detection and Classification of Cardiac Arrhythmias With 12-Lead Electrocardiograms

BackgroundDeep-learning algorithms to annotate electrocardiograms (ECGs) and classify different types of cardiac arrhythmias with the use of a single-lead ECG input data set have been developed. It remains to be determined whether these algorithms can be generalized to 12-lead ECG-based rhythm classification. MethodsWe used a long short-term memory (LSTM) model to detect 12 heart rhythm classes with the use of 65,932 digital 12-lead ECG signals from 38,899 patients, using annotations obtained by consensus of 3 board-certified electrophysiologists as the criterion standard. ResultsThe accuracy of the LSTM model for the classification of each of the 12 heart rhythms was ≥ 0.982 (range 0.982-1.0), with an area under the receiver operating characteristic curve of ≥ 0.987 (range 0.987-1.0). The precision and recall ranged from 0.692 to 1 and from 0.625 to 1, respectively, with an F1 score of ≥ 0.777 (range 0.777-1.0). The accuracy of the model (0.90) was superior to the mean accuracies of internists (0.55), emergency physicians (0.73), and cardiologists (0.83). ConclusionsWe demonstrated the feasibility and effectiveness of the deep-learning LSTM model for interpreting 12 common heart rhythms according to 12-lead ECG signals. The findings may have clinical relevance for the early diagnosis of cardiac rhythm disorders.

1080 Bilirubin and QT Interval Prolongation (From the Third National Health and Nutrition Examination Survey)

INTRODUCTION: The association of bilirubin with cardiovascular disease (CVD) is controversial. We sought to explore the association of total bilirubin levels with QT interval in a multiracial cohort. METHODS: A total of 6,627 participants (59.0 ± 13.3 years; 52.6%% women, 49.7% Non-Hispanic Whites) without cardiovascular disease (CVD) from the Third National Health and Nutrition Examination Survey (NHANES-III) were included in this analysis. QT was automatically measured from digital 12-lead electrocardiogram in a central reading center. A multivariable logistic regression model was used to examine cross-sectional association between tertiles of total bilirubin and prolonged QT interval (≥450 ms in men, ≥460 ms in women). RESULTS: The prevalence of prolonged QT was higher among those with higher levels of total bilirubin (prolonged QT prevalence was 4.7%, 6.8%, and 7.0% across total bilirubin lower (0-0.4 mg/dl), middle (0.5-1.6 mg/dl) and higher (0.70-4.30 mg/dl) tertiles respectively). In a model adjusted for potential confounders, participants within the highest total bilirubin tertile had significantly greater odds of prolonged QT interval (OR (95% CI): 1.53 (1.16-2.02) compared to those with bilirubin levels in the first tertile. Each 0.29 mg/dl increase in total bilirubin levels was associated with a 12% (P-value &lt; .0001) increase in the prevalence of prolonged QT interval (Table 1). This association was stronger in men than women (interaction P-value = 0.04). CONCLUSION:: Recent evidence suggests worse cardiovascular (CV) outcomes with high levels of bilirubin but the mechanism behind worse CV events remains to be elucidated. Our study showed that elevated bilirubin levels are associated with prolonged QT interval. QT interval has been consistently associated with increased risk of all-cause and CV death. Thus, finding an association between total bilirubin levels and prolonged QT interval in our study extends our current knowledge on the relationship between serum bilirubin and CVD by demonstrating a link between higher total bilirubin and abnormal cardiac repolarization. To the best of our knowledge, this is the first study to demonstrate the association between bilirubin and prolonged QT. This association was stronger in men than women. Future studies are needed to establish the causal relationship between bilirubin and QT prolongation and how this association between bilirubin and QT interval could be harnessed to improve CVD risk estimation.

Relation of total bilirubin and QT interval prolongation (from the Third National Health and Nutrition Examination Survey).

The association of bilirubin with cardiovascular disease (CVD) is controversial. We sought to explore the association of total bilirubin (TB) levels with QT interval in a multiracial cohort. A total of 6,627 participants (59.0±13.3years; 52.6% women, 49.7% Non-Hispanic Whites) without CVD from the Third National Health and Nutrition Examination Survey were included in this analysis. QT was automatically measured from digital 12-lead electrocardiogram in a central reading center. A multivariable logistic regression model was used to examine the cross-sectional association between tertiles of TB and prolonged QT interval (≥450ms in men and ≥460ms in women). The prevalence of prolonged QT was higher among those with higher levels of TB (prolonged QT prevalence was 4.7%, 6.8%, and 7.0% across TB lower (0-0.4mg/dl), middle (0.5-1.6mg/dl), and higher (0.70-4.30mg/dl) tertiles, respectively). In a model adjusted for potential confounders, participants within the highest TB tertile had significantly greater odds of the prolonged QT interval (Odds ratios [95% confidence interval] 1.53 [1.16-2.02]) compared to those with bilirubin levels in the first tertile. Each 0.29mg/dl increase in TB levels was associated with a 12% (p-value <.0001) increase in the prevalence of prolonged QT interval. This association was stronger in men than in women (interaction p-value=.04). Elevated bilirubin levels are associated with a prolonged QT interval. This finding extends our current knowledge on the relationship between serum bilirubin and CVD by demonstrating a link between higher TB and abnormal cardiac repolarization.

Relation of Marital Status and QT Interval Prolongation (from the Third National Health and Nutrition Examination Survey)

Although the link between marital status and mortality is well established, the pathophysiological basis is unclear. An investigation of the association of marital status with prolonged QT interval may highlight the underlying mechanism for poor outcomes associated with being unmarried. This analysis included 6,562 participants (mean age 58.6 years, 52% women, 50.1% non-Hispanic whites) without a history of cardiovascular disease from the Third National Health and Nutrition Examination Survey. QT was automatically measured from digital 12-lead electrocardiogram in a central reading center. Marital status was defined by self-report as married and unmarried (never married, divorced/separated or widowed). A multivariable logistic regression model was used to examine cross-sectional association between marital status and prolonged QT interval (≥450 ms in men, ≥460 ms in women). Compared with married, unmarried was associated with 46% higher odds of the prolonged QT interval (odds ratio [OR] 95% confidence interval [95% CI]: 1.46[1.16-1.83]). This association was stronger among men versus women (OR[95% CI]: 1.75[1.27-2.41] vs 1.26[0.92-1.73] respectively; interaction p value = 0.03) and in younger versus older participants (OR [95% CI]: 1.72[1.21-2.42] vs 1.40[1.05-1.88], respectively; interaction p value = 0.002). When the types of unmarried were compared to married, a dose-response relation with prolonged QT was observed with the highest odds in never married followed by divorced/separated, and then widowed. In conclusion, marital status is associated with a prolonged QT interval, especially among men and younger participants. Prolonged QT interval may indicate a biologic substrate through which social isolation defined by unmarried state increases the risk of poor outcomes in the future.

Support vector machine-based assessment of the T-wave morphology improves long QT syndrome diagnosis.

Diagnosing long QT syndrome (LQTS) is challenging due to a considerable overlap of the QTc-interval between LQTS patients and healthy controls. The aim of this study was to investigate the added value of T-wave morphology markers obtained from 12-lead electrocardiograms (ECGs) in diagnosing LQTS in a large cohort of gene-positive LQTS patients and gene-negative family members using a support vector machine. A retrospective study was performed including 688 digital 12-lead ECGs recorded from genotype-positive LQTS patients and genotype-negative relatives at their first visit. Two models were trained and tested equally: a baseline model with age, gender, RR-interval, QT-interval, and QTc-intervals as inputs and an extended model including morphology features as well. The best performing baseline model showed an area under the receiver-operating characteristic curve (AUC) of 0.821, whereas the extended model showed an AUC of 0.901. Sensitivity and specificity at the maximal Youden's indexes changed from 0.694 and 0.829 with the baseline model to 0.820 and 0.861 with the extended model. Compared with clinically used QTc-interval cut-off values (>480 ms), the extended model showed a major drop in false negative classifications of LQTS patients. The support vector machine-based extended model with T-wave morphology markers resulted in a major rise in sensitivity and specificity at the maximal Youden's index. From this, it can be concluded that T-wave morphology assessment has an added value in the diagnosis of LQTS.

A larger vectorcardiographic QRS area is associated with left bundle branch block and good prognosis in patients with cardiac resynchronization therapy

BackgroundThe association between the vectorcardiographic QRS area, bundle branch pattern and clinical long-term prognosis in patients who have undergone cardiac resynchronization therapy (CRT) has been unclear. MethodsWe enrolled 50 consecutive patients who underwent CRT. Vectorcardiograms were constructed from preprocedural digital 12-lead electrocardiograms using the inverse Dower method. The vectorcardiographic QRS area was defined as the root of the sum of the square in the integral between the ventricular deflection curve and the baseline from QRS beginning to end in leads X, Y, and Z. The primary endpoints were total mortality and admission due to heart failure. ResultsThe vectorcardiographic QRS area in left bundle branch block (N = 13), right bundle branch block (N = 13), interventricular conduction delay (N = 11) and pacemaker rhythm (N = 13) were 218 ± 99, 97 ± 44, 90 ± 40, and 131 ± 58 μVs, respectively (ANOVA p < 0.001). During the mean follow-up period of 28 (2–86) months, 13 primary endpoints occurred. We divided patients into two groups: a large QRS area group (QRS area ≥114 μVs, N = 25) and a small QRS area group (QRS area <114 μVs, N = 25) by the median. The large QRS area group had a significantly lower rate of the primary endpoint compared with that of the small QRS area group (log rank 4.35, p = 0.037). The Cox regression analysis revealed that a QRS area <114 μVs was a significant predictor of the primary endpoint (HR 3.98, 95% CI 1.01–15.63, p = 0.048). ConclusionsA larger preprocedural vectorcardiographic QRS area was associated with left bundle branch block and good prognosis in patients who underwent CRT.