12-lead Electrocardiogram Research Articles

Role of rapid 12-lead electrocardiogram in triage initiatives for ST-elevation myocardial infarction patients self-presenting in emergency departments. A systematic review and meta-analysis.

To assess impact of triage initiatives for rapid 12-lead electrocardiogram acquisition on Door-To-ECG, Door-To-Balloon, length of stay, and in-hospital mortality for self-presenting emergency department patients with ST-elevation myocardial infarction. This systematic review encompassed cohort studies, controlled trials, one-group pretest-post-test studies, interventional, observational, and randomized controlled trials assessing rapid acquisition of electrocardiograms for patients above 18 years experiencing symptoms of ST-elevation myocardial infarction in emergency departments. Data from seven databases underwent screening, extraction, and quality appraisals by two independent reviewers. Employing a random-effects model, meta-analyses were conducted for primary outcomes; Door-To-ECG, Door-To-Balloon, length of stay, and in-hospital mortality. Subgroup analyses and meta-regression were performed for meta-analyses with over 10 studies. This review included 25 studies with 19,475 ST-elevation myocardial infarction patients. All were cohort studies with acceptable evidence quality. Our findings revealed that enhanced triage initiatives for electrocardiograms related to significant reductions in Door-To-ECG (MD -6.45 minutes, P < 0.001) and Door-To-Balloon (MD -24.40 minutes, P < 0.001) times. More institutions met benchmarked goals for Door-To-ECG (MD 22.2%, P < 0.001) and Door-To-Balloon (MD 15.6%, P < 0.001) times. Improvements reported in length of stays and in-hospital mortality were not significant. Subgroup and meta-regression analyses revealed significant differences in Door-To-ECG times but not in Door-To-Balloon times. Positive impacts of such initiatives on ST-elevation myocardial infarction patient outcomes offer institutions opportunities to improve triage processes and training. Future research should focus on extended follow-up and larger sample sizes for a comprehensive understanding of sustained impacts. PROSPERO: CRD42023472392.

The Effect of QT, QTc, and Tpe Interval on Outcome and Mortality in Syncope Patients

Abstract Introduction Syncope is a transient condition associated with disturbance of cerebral perfusion. Mortality is higher in cases of syncope of cardiac origin, which is observed in patients with arrhythmogenic findings on electrocardiogram (ECG). In this study, we plan to investigate the relationship of QT interval (QT), Corrected QT Interval (QTc), T peak end (Tpe), Tpe/QT, and Tpe/QTc measured on 12-lead ECGs of patients with mortality and outcomes of patients. Materials and Methods Our study was conducted prospectively with syncope patients older than 18 years presenting to the emergency department of a tertiary teaching and research hospital. The demographic data, ECGs, laboratory data, and outcomes of the syncope patients were recorded and analyzed. Results Of the 80 syncope patients who participated in the study, 36% were women. The V5 interval of the syncope patients was significantly longer than that of the control group. When comparing the TpeV5/QTc rates of the patients, the syncope patients had significantly higher rates. In a categorical comparison, the Tpe times of the syncope patients were more frequently prolonged than those of the control group. Conclusion The measurement of Tpe interval, which is prolonged in patients, appears as a parameter for syncope. The prolongation of the Tpe time observed in lead V5 of the ECG requires a closer follow-up of the patients, especially during the follow-up in daily practice, which represents an important part of the emergency admissions.

Evaluating the incidence of syncope and electrocardiogram changes in methadone-treated volunteers quitting addiction in yazd

Objectives: Methadone is a synthetic opioid with a high affinity for opioid receptors, widely used as an effective intervention for opioid dependence and the management of acute and chronic pain. However, its use has been associated with serious cardiac side effects, including torsades de pointes (TdP) and prolonged QTc intervals. This study investigated the incidence of syncope and electrocardiogram (ECG) changes in individuals who underwent methadone treatment for addiction. Methods: This cross-sectional study, conducted in Yazd in 2020, involved 100 participants seeking methadone treatment at an addiction clinic. A cardiologist recorded initial ECGs using a 12-lead ECG device, assessing parameters such as heart rhythm, QTc interval, and premature contractions. Participants were informed about the study procedures and the necessity of follow-up ECGs one week and three months after starting methadone. The researchers monitored the patients for three months with regular outpatient visits. Results: The findings revealed a significant difference in syncope incidence based on ECG rhythm on day seven and month three, with higher rates observed in patients exhibiting junctional rhythms or PVC (p-value=0.000). Additionally, a significant relationship was found between syncope frequency and U wave status at month three (p-value=0.046).No significant associations were noted for ST segment changes or T wave variations. Furthermore, higher methadone doses, longer QTc intervals on day seven, and prolonged PR intervals correlated with increased syncope occurrences. Conclusions: This study indicates that ECG-related variables and methadone dosage significantly influence syncope incidence among treated patients. Specifically, individuals with junctional rhythms or PVC and those receiving higher doses of methadone were at greater risk for syncope. Monitoring PR and QTc intervals is crucial for preventing syncope in patients undergoing methadone treatment.

Functional feature extraction and validation from twelve-lead electrocardiograms to identify atrial fibrillation

BackgroundDeep learning methods on standard, 12-lead electrocardiograms (ECG) have resulted in the ability to identify individuals at high-risk for the development of atrial fibrillation. However, the process remains a “black box” and does not help clinicians in understanding the electrocardiographic changes at an individual level. we propose a nonparametric feature extraction approach to identify features that are associated with the development of atrial fibrillation (AF).MethodsWe apply functional principal component analysis to the raw ECG tracings collected in the Chronic Renal Insufficiency Cohort (CRIC) study. We define and select the features using ECGs from participants enrolled in Phase I (2003–2008) of the study. Cox proportional hazards models are used to evaluate the association of selected ECG features and their changes with the incident risk of AF during study follow-up. The findings are then validated in ECGs from participants enrolled in Phase III (2013–2015).ResultsWe identify four features that are related to the P-wave amplitude, QRS complex and ST segment. Both their initial measurement and 3-year changes are associated with the development of AF. In particular, one standard deviation in the 3-year decline of the P-wave amplitude is independently associated with a 29% increased risk of incident AF in the multivariable model (HR: 1.29, 95% CI: [1.16, 1.43]).ConclusionsCompared with deep learning methods, our features are intuitive and can provide insights into the longitudinal ECG changes at an individual level that precede the development of AF.

ECG-surv: A deep learning-based model to predict time to 1-year mortality from 12-lead electrocardiogram

Electrocardiogram (ECG) abnormalities have demonstrated potential as prognostic indicators of patient survival. However, the traditional statistical approach is constrained by structured data input, limiting its ability to fully leverage the predictive value of ECG data in prognostic modeling. This study aims to introduce and evaluate a deep-learning model to simultaneously handle censored data and unstructured ECG data for survival analysis. We herein introduce a novel deep neural network called ECG-surv, which includes a feature extraction neural network and a time-to-event analysis neural network. The proposed model is specifically designed to predict the time to 1-year mortality by extracting and analyzing unique features from 12-lead ECG data. ECG-surv was evaluated using both an independent test set and an external set, which were collected using different ECG devices. The performance of ECG-surv surpassed that of the Cox proportional model, which included demographics and ECG waveform parameters, in predicting 1-year all-cause mortality, with a significantly higher concordance index (C-index) in ECG-surv than in the Cox model using both the independent test set (0.860 [95% CI: 0.859-0.861] vs. 0.796 [95% CI: 0.791-0.800]) and the external test set (0.813 [95% CI: 0.807-0.814] vs. 0.764 [95% CI: 0.755-0.770]). ECG-surv also demonstrated exceptional predictive ability for cardiovascular death (C-index of 0.891 [95% CI: 0.890-0.893]), outperforming the Framingham risk Cox model (C-index of 0.734 [95% CI: 0.715-0.752]). ECG-surv effectively utilized unstructured ECG data in a survival analysis. It outperformed traditional statistical approaches in predicting 1-year all-cause mortality and cardiovascular death, which makes it a valuable tool for predicting patient survival.

Screening and Interpretation Rules for Resting Electrocardiograms in Polish Air Force Aircrew

INTRODUCTION: The resting 12-lead electrocardiogram (ECG) is one of the basic screening tests used in assessing fitness for military service and in periodic examinations of professional soldiers. Its purpose is to reduce the risk of death from cardiovascular disease by early detection of cardiac pathologies predisposing to sudden cardiac arrest. Commonly used electrocardiographic standards do not perform well in screening, accounting for the recognition of a large number of abnormalities unrelated to structural heart disease. METHODS: A comprehensive literature review was performed using electronic databases such as PubMed®, Google Scholar, and relevant aerospace medical journals. The search focused on keywords, including “screening ECG,” “electrocardiographic criteria,” “military aircrew,” and “aviation medicine.” Articles that discussed ECG interpretation rules and the implementation of screening protocols in military aircrew were included. The review also involved an examination of guidelines and policies from major aviation regulatory bodies. RESULTS: This review includes information on topics relevant to the application of 12-lead ECG in screening diagnostics, such as general ECG interpretation criteria, limitations of ECG as a screening tool, objectives of ECG-based screening, and past experiences with the use of screening ECG in athletes within the context of military medicine. DISCUSSION: We discuss various criteria for the interpretation of screening ECG in athletes and soldiers to increase the specificity of this test. We also present a proposal for a system of criteria for interpreting ECGs in aircrew of the Polish Air Force, which may help improve the detection of cardiac pathologies in this professional group. Kurek MA, Rola M. Screening and interpretation rules for resting electrocardiograms in Polish Air Force aircrew. Aerosp Med Hum Perform. 2025; 96(2):135–142.

Corticosteroid-Induced Sinus Bradycardia in a Dog with Systemic Lupus Erythematosus: A Case Report

A 3-year-old, 10.2 kg, spayed, female mixed-breed dog was diagnosed with systemic lupus erythematosus. The patient was administered intravenous methylprednisolone at a dose of 2 mg/kg twice daily. Over the next 24 h, the dog’s heart rate dropped to 42–48 bpm, while other vital signs remained stable. A 12-lead electrocardiogram revealed sinus bradycardia without any other arrhythmia. Notably, the dog was not receiving any beta-blockers or undergoing other therapies commonly associated with sinus bradycardia. After tapering the corticosteroid dose, no further episodes of bradycardia were observed. Herein, we report the first case of sinus bradycardia in a dog following intravenous methylprednisolone therapy to improve veterinarians’ understanding of corticosteroid-induced cardiac arrhythmias.

Association between blood test indicators and atrial fibrillation in elderly patients aged 65 and above in the Central Jiangsu region: a cross-sectional study

ObjectivesThe aim of this study was to explore the correlation between blood test indicators and Atrial Fibrillation (AF) in Individuals Aged 65 and Older in Yangzhou, Jiangsu.MethodsFrom January 1, 2019, to August 31, 2023, an epidemiological cross-sectional survey was conducted among the elderly population undergoing health check-ups at Northern Jiangsu People’s Hospital in Jiangsu Province. Patients diagnosed with AF after a 12-lead electrocardiogram were included in the case group, and non-AF individuals matched by age and gender in a 1:4 frequency ratio were included in the control group. Least Absolute Shrinkage and Selection Operator (LASSO) regression was used to select important variables from routine blood tests and biochemical indicators and their derived indicators (such as TyG, TyG-BMI, TG/HDL-c, RAR, NLR, MHR). Based on the selected variables, participants were divided into four groups (Q1 ~ Q4), and multifactorial Logistic regression analysis, restricted cubic spline regression, gender subgroup analysis and Receiver Operating Characteristic (ROC) curve were used to explore the relationship between the relevant variables and AF.ResultsA total of 5,879 elderly individuals over the age of 65 were included in the study, with a prevalence of AF of 2.96% (174/5,879). The prevalence of AF in the overall population, as well as in male and female populations, showed a continuous increasing trend with age (P for trend < 0.001). A total of 696 individuals without AF after matching served as the control group, and LASSO regression identified albumin, direct bilirubin, and uric acid as three significant indicators. After adjusting for relevant confounding factors, lower levels of albumin, and higher levels of direct bilirubin and uric acid were significantly associated with the occurrence of AF (P < 0.05). Gender subgroup analysis revealed that in the elderly female population, albumin was not significantly associated with AF (P > 0.05), while direct bilirubin and uric acid were significantly associated with AF (P < 0.05). In the male population, albumin, direct bilirubin, and uric acid were significantly associated with AF (P < 0.05). Restricted cubic spline regression analysis showed a significant nonlinear relationship between direct bilirubin and AF (P for nonlinear < 0.001). The ROC curve analysis indicates that albumin, direct bilirubin, and uric acid all have good association strength with AF in elderly patients, with direct bilirubin showing the strongest association effect (AUC (95% CI) = 0.728 (0.686, 0.769)).ConclusionsLow levels of albumin, high levels of direct bilirubin, and uric acid are all significantly associated with AF in the elderly population of the Central Jiangsu region. The conclusions of this study need further validation with a larger sample size.

Determining the QRS axis: visual estimation is equal to calculation.

The QRS axis of the electrocardiogram (ECG) is often considered in clinical practice, but its determination is frequently limited to arough estimation, such as "normal", with left or right deviation, and superior or inferior in the case of premature ventricular complexes (PVCs). However, amore exact determination of the QRS axis may be warranted in certain scenarios, such as to determine the origin of PVCs more precisely, and is attainable by visual estimation using the hexaxial reference system. The aim of this study was to determine how well such an estimation of the QRS axis would correlate with the axis calculated by formulas. APVC database from 2012-2020 was used to extract 12-lead ECGs of patients with outflow tract PVCs and analyze the QRS axes of regularly conducted beats as well as PVCs. QRS axes were determined visually by two physicians with the help of the hexaxial reference system to an accuracy of10° on the one hand, and were calculated using three previously described formulas based on QRS voltages on the other. Atotal of 216 QRS complexes from 108 patients were analyzed (108 regularly conducted beats and 108 PVCs). Estimated QRS axes of regularly conducted beats and PVCs were 39 ± 40° and 88 ± 15°, respectively. Calculated QRS axes of regularly conducted beats according to the three formulas were 37 ± 40°, 36 ± 40° and 35 ± 38°, respectively. Calculated QRS axes of PVCs according to the three formulas were 87 ± 15°, 87 ± 14° and 86 ± 16°, respectively. Correlation coefficients showed strong correlations between the estimated and the three calculated values for regularly conducted beats (0.98, 0.97 and 0.98) and PVCs (0.94, 0.94 and 0.94). Asophisticated visual estimation of the QRS axis correlates well with voltage-based calculations and can therefore be considered sufficient for most purposes.

Comparing the Effects of Square, 4-7-8, and 6 Breaths-per-Minute Breathing Conditions on Heart Rate Variability, CO2 Levels, and Mood.

Square and 4-7-8 breathing are popularly promoted by psychotherapists but have little empirical support. We hypothesized that breathing at 6 breaths per minute (bpm) would improve HRV, reduce blood pressure, and improve mood more than either square or 4-7-8 breathing. We also hypothesized square and 4-7-8 breathing would increase end-tidal CO2 (PETCO2). Eighty-four college students (60% female) had HRV measured at baseline and during 1) square breathing, 2) 4-7-8 breathing, 3) 6bpm with a 4:6 inhale-to-exhale ratio, and 4) 6bpm with a 5:5 ratio. We used a 3-lead electrocardiogram to measure HRV and a capnometer to measure PETCO2. We measured mood using the Scale of Positive and Negative Experience (SPANE) after each breathing condition. We used a repeated measures mixed effects model to examine the effects of each breathing technique on HRV, blood pressure, mood, and PETCO2 levels. Breathing at 6bpm increased HRV measures more than square or 4-7-8 breathing with small to medium effects. None of the breathing conditions resulted in meaningful changes in blood pressure or mood. Unexpectedly, breathing at 6bpm led to mild over-breathing. While various breathing techniques are used in psychotherapy, their relative efficacy remains unclear. This study found that breathing at 6bpm is likely more effective at increasing HRV than square or 4-7-8 breathing, albeit with a higher risk of over-breathing. Future studies could examine whether these effects hold over multiple weeks of breathing practice and attend to the effects of different breathing techniques on CO2 levels.

ELECTROCARDIOGRAPHY FINDINGS IN BLUNT CHEST TRAUMA AT TERTIARY CARE HOSPITAL PESHAWAR

Introduction: Blunt chest trauma is a significant cause of morbidity and mortality, often leading to a range of cardiac complications detectable through electrocardiogram (ECG) abnormalities. Objectives: The primary objective of this research was to identify the early onset of myocardial damage in patients with acute blunt chest injuries using ECG. The secondary objective is being to compare the ECG changes between paediatric and adult trauma patients. Methodology: This cross-sectional study was conducted in the Emergency department (ED) and patients admitted with acute blunt chest trauma were included in the current study. A total of 281 patients were included based on the following inclusion criteria: patients of all ages with blunt chest trauma. Exclusion criteria included preexisting cardiac and neurological conditions, penetrating chest injuries, pregnant women and incomplete medical records. Demographic and clinical data were collected, and a standard 12-lead ECG was performed and analyzed for abnormalities. Statistical analysis was done using SPSS version 26, where chi square test was applied to identify factors that may be associated with ECG abnormalities and assess age-related differences in ECG findings. Results: ECG findings in patients with blunt chest trauma, recorded in 30.9% of the total sample. Sinus arrhythmia's, particularly sinus tachycardia were the most common findings in both adults (6.2%) and pediatrics (8.1%). In adult’s sinus tachycardia was followed by ST segment changes and atrial fibrillation. In pediatrics, sinus tachycardia was accompanied by PVCs and supraventricluar tachycardia. Conclusion: Sinus arrhythmias were the most common findings in both group, followed by ST segment changes and atrial fibrillation in adults while in pediatrics, followed by PVCs and SVT.

Machine Learning-Based Clustering Using a 12-Lead Electrocardiogram in Patients With a Implantable Cardioverter Defibrillator to Identify Future Ventricular Arrhythmia

Implantable cardioverter defibrillators (ICDs) reduce mortality associated with ventricular arrhythmia in high-risk patients with cardiovascular disease. Machine learning (ML) approaches are promising tools in arrhythmia research; however, their application in predicting ventricular arrhythmias in patients with ICDs remains unexplored. We aimed to predict and stratify ventricular arrhythmias requiring ICD therapy using 12-lead electrocardiograms (ECGs) in patients with an ICD. This retrospective analysis included 200 adult patients who underwent ICD implantation at a single center. Patient demographics, clinical features, and 12-lead ECG data were collected. Unsupervised learning techniques, including K-means and hierarchical clustering, were used to stratify patients based on 12-lead ECG features. Dimensionality reduction methods were also used to optimize clustering accuracy. The silhouette coefficient was used to determine the optimal method and number of clusters. Of the 200 patients, 59 (29.5%) received appropriate therapy. The mean age of patients was 62.3 years, and 81.0% were male. The mean follow-up period was 2,953 days, with no significant intergroup differences. Hierarchical clustering into 3 clusters proved to be the most accurate (silhouette coefficient=0.585). Kaplan-Meier curves for these 3 clusters revealed significant differences (P=0.026). We highlight the potential of ML-based clustering using 12-lead ECGs to help in the risk stratification of ventricular arrhythmia. Future research in a larger multicenter setting may provide further insights and refine ICD indications.

Usefulness of Apple Watch Electrocardiograms for Identifying the Origin of Premature Ventricular Contractions

The Apple Watch (AW) can record single-lead electrocardiograms (ECGs) and has been investigated for arrhythmia detection. In this study we evaluated its accuracy in identifying the origin of premature ventricular contractions (PVCs) vs. standard 12-lead ECGs. A total of 7 patients with PVCs were assessed using both 12-lead and AW ECG recordings. The QRS polarity observed in the AW recordings was consistent with that of the standard ECGs in most cases, demonstrating its utility in estimating three distinct PVC origins. The AW holds potential as an auxiliary tool for PVC origin assessment, contributing to arrhythmia management in clinical practice.

Differential diagnostic value of P wave dispersion and QT interval dispersion between psychogenic pseudosyncope and vasovagal syncope in children and adolescents

BackgroundBoth psychogenic pseudosyncope (PPS) and vasovagal syncope (VVS) in children and adolescents are diseases of transient loss of consciousness. It is difficult to distinguish them clinically. This paper will study the differential diagnostic value of P wave dispersion (Pd) and QT interval dispersion (QTd) between PPS and VVS.MethodsThe 31 children with PPS and 40 children with VVS from July 2014 to November 2023 were enrolled as the study group. Meanwhile, 30 healthy children who underwent a physical examination at the same hospital were matched to the control group. P wave duration and QT interval in the 12-lead electrocardiogram were measured at the baseline.Results(1) Comparison between groups: ① The Pd, corrected P wave dispersion (Pcd), QTd, and corrected QT interval dispersion (QTcd) in PPS group were significantly higher than those in control group (P < 0.05). The minimum P wave duration (Pmin) and corrected P wave duration (Pcmin) in PPS group were significantly lower than those in control group (P < 0.05). There were no significant differences in maximum P wave duration (Pmax), corrected maximum P-wave duration (Pcmax), maximum QT interval (QTmax), minimum QT interval (QTmin), corrected maximum QT interval (QTcmax), and corrected minimum QT interval (QTcmin) between PPS group and control group (P > 0.05). ② The Pd, Pcd, QTd, and QTcd in VVS group were significantly higher than those in control group (P < 0.05). The Pmin, Pcmin, and QTcmin in VVS group were significantly lower than those in control group (P < 0.05). There were no significant differences in Pmax, Pcmax, QTmax, QTmin, and QTcmax between VVS group and control group (P > 0.05). ③ The Pmax, Pd, QTmax, QTd, and QTcd in PPS group were significantly lower than those in VVS group (P < 0.05). There were no significant differences in Pmin, Pcmax, Pcmin, Pcd, QTmin, QTcmax, and QTcmin between PPS group and VVS group (P > 0.05). (2) ROC curve: Pmax, Pd, QTmax, QTd, and QTcd had a certain differential diagnostic value between PPS and VVS in children and adolescents (P < 0.05). QTd had the largest area under curve (0.735), with a sensitivity of 85.00% and a specificity of 53.30% at the cut off value of ≥ 28.11 ms for VVS diagnosis.ConclusionsIn children and adolescents, electrocardiogram parameters such as Pmax, Pd, QTmax, QTd, and QTcd all possess predictive value in differentiating between PPS and VVS. Among them, QTd has the greatest differential diagnostic value.

Inflammatory Biomarkers and Lipid Parameters May Predict an Increased Risk for Atrial Arrhythmias in Patients with Systemic Sclerosis.

Autoimmune inflammation enhances the electrical instability of the atrial myocardium in patients with systemic sclerosis (SSc); thus, atrial arrhythmia risk is increased, which might be predicted by evaluating the P wave interval and dispersion of a 12-lead surface electrocardiogram (ECG). We examined 26 SSc patients and 36 healthy controls and measured the P wave interval and P wave dispersion of the 12-lead surface ECG in each patient. Furthermore, echocardiography and 24-h Holter ECG were performed and levels of inflammatory laboratory parameters, including serum progranulin (PGRN), sVCAM-1, sICAM-1, leptin and C-reactive protein (CRP), were determined. Lipid parameters, such as Apo A-I, LDL-cholesterol (LDL-C), oxidized LDL (oxLDL) and the LDL and HDL subfractions were also evaluated. The P wave interval showed a significant positive correlation with the levels of Apo A-I, LDL-C, CRP, sVCAM-1, sICAM-1 and leptin. The oxLDL level correlated positively with P wave dispersion. Of note, significant positive correlation was also found between the large HDL percentage and the P wave interval. Our results suggest that PGRN, sVCAM-1, sICAM-1, leptin, CRP, LDL-C and oxLDL, along with LDL and HDL subfractions, might have a role in atrial arrhythmogenesis in patients with SSc.

A Brief History of Clinical Electrocardiography: A Century After Einthoven's Nobel Prize.

The invention of the string galvanometer by Willem Einthoven in 1901 ushered in a new era of clinical investigation. Previous instruments were capable only of rudimentary measurements that were of limited utility. Advances in physiology and engineering allowed Einthoven to construct a device that was uniquely capable of recording the minute electrical currents of the heart. Early string galvanometers were massive, immovable machines. Over time, the apparatus became smaller and portable, allowing examinations at the bedside. In the decades that followed, clinicians used the now ubiquitous instrument to evaluate cardiac arrhythmias, coronary artery disease, and conduction abnormalities. The remainder of the century saw the evolution of the contemporary 12-lead electrocardiogram, with standards established regarding technique and nomenclature. Awarded the Nobel Prize in 1924 for "his discovery of the mechanism of the electrocardiogram," Einthoven's legacy endures in modern clinical medicine.

Electrocardiogram-based machine learning for risk stratification of patients with suspected acute coronary syndrome.

The importance of risk stratification in patients with chest pain extends beyond diagnosis and immediate treatment. This study sought to evaluate the prognostic value of electrocardiogram feature-based machine learning models to risk-stratify all-cause mortality in those with chest pain. This was a prospective observational cohort study of consecutive, non-traumatic patients with chest pain. All-cause death was ascertained from multiple sources, including the CDC National Death Index registry. Six machine learning models were trained for survival analysis using 73 morphological electrocardiogram features (80% training with 10-fold cross-validation and 20% testing), followed by a variational Bayesian Gaussian mixture model to define distinct risk groups. The resulting classification performance was compared against the HEART score. The derivation cohort included 4015 patients (age 59 ± 16 years, 47% women). The mortality rate was 20.3% after a median follow-up period of 3.05 years (interquartile range 1.75-5.32). Extra Survival Trees outperformed other forecasting models, and the derived risk groups successfully classified patients into low-, moderate-, and high-risk groups (log-rank test statistic = 121.14, P < .001). This model outperformed the HEART score, reducing the rate of missed events by >90% with a negative predictive value and sensitivity of 93.4% and 85.9%, compared to 89.0% and 75.0%, respectively. In an independent external testing cohort (N = 3095, age 59 ± 15 years, 44% women, 30-day mortality 3.5%), patients in the moderate [odds ratio 3.62 (1.35-9.74)] and high [odds ratio 6.12 (2.38-15.75)] risk groups had significantly higher odds of mortality compared to those in the low-risk group. The externally validated machine learning-based model, exclusively utilizing features from the 12-lead electrocardiogram, outperformed the HEART score in stratifying the mortality risk of patients with acute chest pain. This may have the potential to impact the precision of care delivery and the allocation of resources to those at highest risk of adverse events.

Explainable AI associates ECG aging effects with increased cardiovascular risk in a longitudinal population study

Aging affects the 12-lead electrocardiogram (ECG) and correlates with cardiovascular disease (CVD). AI-ECG models estimate aging effects as a novel biomarker but have only been evaluated on single ECGs—without utilizing longitudinal data. We validated an AI-ECG model, originally trained on Brazilian data, using a German cohort with over 20 years of follow-up, demonstrating similar performance (r2 = 0.70) to the original study (0.71). Incorporating longitudinal ECGs revealed a stronger association with cardiovascular risk, increasing the hazard ratio for mortality from 1.43 to 1.65. Moreover, aging effects were associated with higher odds ratios for atrial fibrillation, heart failure, and mortality. Using explainable AI methods revealed that the model aligns with clinical knowledge by focusing on ECG features known to reflect aging. Our study suggests that aging effects in longitudinal ECGs can be applied on population level as a novel biomarker to identify patients at risk early.

Heart failure risk stratification using artificial intelligence applied to electrocardiogram images: a multinational study.

Current heart failure (HF) risk stratification strategies require comprehensive clinical evaluation. In this study, artificial intelligence (AI) applied to electrocardiogram (ECG) images was examined as a strategy to predict HF risk. Across multinational cohorts in the Yale New Haven Health System (YNHHS), UK Biobank (UKB), and Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), individuals without baseline HF were followed for the first HF hospitalization. An AI-ECG model that defines cross-sectional left ventricular systolic dysfunction from 12-lead ECG images was used, and its association with incident HF was evaluated. Discrimination was assessed using Harrell's C-statistic. Pooled cohort equations to prevent HF (PCP-HF) were used as a comparator. Among 231 285 YNHHS patients, 4472 had primary HF hospitalizations over 4.5 years (inter-quartile range 2.5-6.6). In UKB and ELSA-Brasil, among 42 141 and 13 454 people, 46 and 31 developed HF over 3.1 (2.1-4.5) and 4.2 (3.7-4.5) years. A positive AI-ECG screen portended a 4- to 24-fold higher risk of new-onset HF [age-, sex-adjusted hazard ratio: YNHHS, 3.88 (95% confidence interval 3.63-4.14); UKB, 12.85 (6.87-24.02); ELSA-Brasil, 23.50 (11.09-49.81)]. The association was consistent after accounting for comorbidities and the competing risk of death. Higher probabilities were associated with progressively higher HF risk. Model discrimination was 0.718 in YNHHS, 0.769 in UKB, and 0.810 in ELSA-Brasil. In YNHHS and ELSA-Brasil, incorporating AI-ECG with PCP-HF yielded a significant improvement in discrimination over PCP-HF alone. An AI model applied to a single ECG image defined the risk of future HF, representing a digital biomarker for stratifying HF risk.

Unsupervised deep learning of electrocardiograms enables scalable human disease profiling

The 12-lead electrocardiogram (ECG) is inexpensive and widely available. Whether conditions across the human disease landscape can be detected using the ECG is unclear. We developed a deep learning denoising autoencoder and systematically evaluated associations between ECG encodings and ~1,600 Phecode-based diseases in three datasets separate from model development, and meta-analyzed the results. The latent space ECG model identified associations with 645 prevalent and 606 incident Phecodes. Associations were most enriched in the circulatory (n = 140, 82% of category-specific Phecodes), respiratory (n = 53, 62%) and endocrine/metabolic (n = 73, 45%) categories, with additional associations across the phenome. The strongest ECG association was with hypertension (p < 2.2×10-308). The ECG latent space model demonstrated more associations than models using standard ECG intervals, and offered favorable discrimination of prevalent disease compared to models comprising age, sex, and race. We further demonstrate how latent space models can be used to generate disease-specific ECG waveforms and facilitate individual disease profiling.