Search

Brandon Tejada,1,2 Roby Joehanes,1,2 Shih-Jen Hwang,1,2 Tianxiao Huan,1– 3 Chen Yao,1,2 Jennifer E Ho,4 Daniel Levy1,2,5 1Framingham Heart Study, Framingham, MA, USA; 2Population Sciences Branch, National Heart, Lung, and Blood Institute, Bethesda, MD, USA; 3Ophthalmology and Visual Sciences, University of Massachusetts Medical School, Worcester, MA, USA; 4CardioVascular Institute and Division of Cardiology, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, MA, Unites States; 5Boston University School of Medicine, Boston, MA, USACorrespondence: Daniel Levy, Framingham Heart Study, 73 Mt. Wayte Avenue, Suite 2, Framingham, MA, 01702, USA, Email Levyd@nih.govPurpose: Assessing an individual’s systemic inflammatory state is vital to understand inflammation’s role in cardiometabolic diseases and identify those at the greatest risk of disease. We generated global inflammation scores and investigated their associations with cardiometabolic risk factors and adverse outcomes.Patients and Methods: Aggregate Inflammation Scores (AIS) and Principal Component Analysis (PCA) scores were generated for 7287 Framingham Heart Study participants using up to 26 inflammation-related proteins, with higher scores reflecting a pro-inflammatory milieu. Multivariable regression and proportional hazards analyses were conducted to investigate the associations of inflammation with cardiometabolic risk factors and outcomes. The primary outcomes for cross-sectional analyses included age, cigarette smoking, fasting lipid and glucose levels, blood pressure, body mass index (BMI), and hypertension, diabetes, and obesity. For prospective analyses, new-onset hypertension, diabetes, obesity, cardiovascular disease and all-cause mortality were investigated.Results: Higher inflammation scores were associated with smoking and older age, higher BMI, systolic blood pressure, lipids, and glucose levels, and with greater odds of hypertension and diabetes after adjusting for age, sex, cohort, and BMI (all p < 0.001). Higher baseline scores were associated with greater odds of new-onset hypertension after adjusting for traditional risk factors (OR [95% CI] per one standard deviation [1-SD] increase, AIS: 1.33 [1.21– 1.47], PCA score: 1.26 [1.12– 1.42], p < 0.001). The AIS also was associated with new-onset diabetes (1.32 [1.14– 1.52], p < 0.001). Proportional hazards analyses revealed greater risk of new-onset cardiovascular disease events and all-cause mortality (HR [95% CI] per 1-SD, AIS: 1.25 [1.14– 1.37] and 1.32 [1.23– 1.42], PCA score: 1.22 [1.13– 1.33] and 1.40 [1.31– 1.49], p < 0.001).Conclusion: Global inflammation scores encompassing an array of pro- and anti-inflammatory proteins and pathways may enhance risk assessment for cardiometabolic diseases. The AIS and PCA scores provide further opportunities to investigate the mechanisms of inflammation-related risk of disease.Keywords: inflammation scores, cardiovascular disease, heart failure, pathways

Observational and trial data have revealed significant improvement in cardiogenic shock (CS) mortality due to acute myocardial infarction (AMI) after introducing early coronary revascularization. Less is known about CS mortality due to heart failure (HF), which is increasingly recognized as a distinct entity from AMI-CS.In this nationwide observational study, the CDC WONDER database was used to identify national trends in age-adjusted mortality rates (AAMR) due to CS (HF vs. AMI related) per 100,000 people aged 35-84. AAMR from AMI-CS decreased significantly from 1999 to 2009 (AAPC: -6.9% [95%CI -7.7, -6.1]) then stabilized from 2009 to 2020. By contrast, HF-CS associated AAMR rose steadily from 2009 to 2020 (AAPC: 13.3% [95%CI 11.4,15.2]). The mortality rate was almost twice as high in males compared to females in both AMI-CS and HF-CS throughout the study period. HF-CS mortality in the non-Hispanic Black population is increasing more quickly than that of the non-Hispanic White population (AAMR in 2020: 4.40 vs. 1.97 in 100,000). The AMI-CS mortality rate has been consistently higher in rural than urban areas (30% higher in 1999 and 28% higher in 2020).These trends highlight the fact that HF-CS and AMI-CS represent distinct clinical entities. While mortality associated with AMI-CS has primarily declined over the last two decades, the mortality related to HF-CS has increased significantly, particularly over the last decade, and is increasing rapidly among individuals younger than 65. Accordingly, a dramatic change in the demographics of CS patients in modern intensive care units is expected.

The "heterozygote advantage" hypothesis has been postulated regarding the role of human leukocyte antigen (HLA) in non-Hodgkin lymphoma (NHL), where homozygous loci are associated with an increased risk of disease. In this retrospective study, we analyzed the HLA homozygosity of 3789 patients with aplastic anemia (AA), acute lymphocytic leukemia (ALL), acute myeloblastic leukemia (AML), chronic lymphocytic leukemia (CLL), chronic myeloid leukemia (CML), myelodysplastic syndrome (MDS), multiple myeloma (MM), and non-Hodgkin lymphoma (NHL) at HLA-A, B, C, DRB1 and DQB1 loci compared to 169,964 normal controls. HLA homozygosity at one or more loci was only associated with an increased risk in NHL patients (OR = 1.28, 95% CI [1.09, 1.50], p = 0.002). This association was not seen in any of the other hematologic diseases. Homozygosity at HLA-A alone, HLA-B + C only, and HLA-DRB1 + DQB1 only was also significantly associated with NHL. Finally, we observed a 17% increased risk of NHL with each additional homozygous locus (OR per locus = 1.17, 95% CI [1.08, 1.25], p trend = 2.4 × 10

Postexercise heart rate recovery (HRR) is an important indicator of cardiac autonomic function and abnormal HRR is associated with adverse outcomes. We hypothesized that deep learning on resting electrocardiogram (ECG) tracings may identify individuals with impaired HRR.We trained a deep learning model (convolutional neural network) to infer HRR based on resting ECG waveforms (HRRAmong 56,793 individuals (mean age 57 years, 51% women), the HRRDeep learning-derived estimates of HRR using resting ECG independently associated with future clinical outcomes, including new-onset DM and all-cause mortality. Inferring postexercise heart rate response from a resting ECG may have potential clinical implications and impact on preventive strategies warrants future study.

The left ventricular outflow tract (LVOT) and ascending aorta are spatially complex, with distinct pathologies and embryologic origins. Prior work examined the genetics of thoracic aortic diameter in a single plane.We sought to elucidate the genetic basis for the diameter of the LVOT, aortic root, and ascending aorta.Using deep learning, we analyzed 2.3 million cardiac magnetic resonance images from 43,317 UK Biobank participants. We computed the diameters of the LVOT, the aortic root, and at 6 locations of ascending aorta. For each diameter, we conducted a genome-wide association study and generated a polygenic score. Finally, we investigated associations between these scores and disease incidence.A total of 79 loci were significantly associated with at least 1 diameter. Of these, 35 were novel, and most were associated with 1 or 2 diameters. A polygenic score of aortic diameter approximately 13 mm from the sinotubular junction most strongly predicted thoracic aortic aneurysm (n = 427,016; mean HR: 1.42 per SD; 95% CI: 1.34-1.50; P = 6.67 × 10We detected distinct genetic loci underpinning the diameters of the LVOT, aortic root, and at several segments of ascending aorta. We spatially defined a region of aorta whose genetics may be most relevant to predicting thoracic aortic aneurysm. We further described a genetic signature that may predispose to aortic stenosis. Understanding genetic contributions to proximal aortic diameter may enable identification of individuals at risk for aortic disease and facilitate prioritization of therapeutic targets.

This study aims firstly to compare the complaints of local native-Malay speakers and expatriate native-English speakers in Brunei in terms of move structure and levels of directness combined with the frequency of modality markers; and secondly, it attempts to address the relationship between polite behaviour and its effectiveness in eliciting the appropriate response from the hearer. Data from an oral discourse completion task show interesting similarities and differences in the complaint move structure between the two groups of speakers. Superficially, there appears to be no significant difference between the two sets of complaints in terms of levels of directness, but a detailed analysis shows each group employing different mitigating strategies to minimise the force of a complaint. Furthermore, responses from an acceptability judgement questionnaire indicate that being indirect, and therefore polite, may not be effective in eliciting the appropriate response to a request for action in a complaint speech act.

ObjectivesPaediatric patients with tracheostomies are a vulnerable group. During the coronavirus disease 2019 pandemic, healthcare workers can be anxious about viral transmission from secretions and aerosols emerging from the open airway. This paper aims to share a systematic approach to decrease staff exposure and optimise care of these patients.MethodsThree documents were developed: a generic tracheostomy management plan detailing troubleshooting; a personalised management plan with customised recommendations; and a guide for tracheostomy tube change to minimise aerosol production.ResultsThe plan was distributed to 31 patients (age range, 11 months to 17 years) including 23 (74.2 per cent) with uncuffed tubes and 9 (29 per cent) on long-term ventilation. There have been 10 occasions in which the plan was utilised and influenced management.ConclusionA structured approach to emergency presentations during the coronavirus disease 2019 pandemic may safeguard paediatric patients from unnecessary manipulation of their tracheostomy tube, minimise viral exposure and allow provision of expeditious care.

OBJECTIVE: To determine the effect of diet, exercise, and diet-exercise in combination on measures of biological age. DESIGN: Secondary analysis of a 1-year randomized, controlled trial SETTING: University-based Medical Center PARTICIPANTS: One-hundred-seven older (age≥65 yrs.) adults with obesity (BMI≥30 kg/m(2)) were randomized and 93 completed the study. Analyses used intention-to-treat. INTERVENTIONS: Participants were randomized to a control group, a weight-management (diet) group, an exercise group, or a weight-management-plus-exercise (diet-exercise group). MAIN OUTCOME MEASURES: We calculated Klemera-Doubal Method (KDM) biological age, Homeostatic Dysregulation (HD) score, and Health Aging Index (HAI) score at baseline, and changes at 6- and 12-months. RESULTS: Diet and diet-exercise decreased KDM biological age more than exercise and control (−2.4±0.4, −2.2±0.3, −0.2±0.4, and 0.2±0.5, respectively, P

Valvular heart disease is associated with a high global burden of disease. Even mild aortic stenosis confers increased morbidity and mortality, prompting interest in understanding normal variation in valvular function at scale.We developed a deep learning model to study velocity-encoded magnetic resonance imaging in 47,223 UK Biobank participants. We calculated eight traits, including peak velocity, mean gradient, aortic valve area, forward stroke volume, mitral and aortic regurgitant volume, greatest average velocity, and ascending aortic diameter. We then computed sex-stratified reference ranges for these phenotypes in up to 31,909 healthy individuals. In healthy individuals, we found an annual decrement of 0.03cm2in the aortic valve area. Participants with mitral valve prolapse had a 1 standard deviation [SD] higher mitral regurgitant volume (P=9.6 × 10−12), and those with aortic stenosis had a 4.5 SD-higher mean gradient (P=1.5 × 10−431), validating the derived phenotypes’ associations with clinical disease. Greater levels of ApoB, triglycerides, and Lp(a) assayed nearly 10 years prior to imaging were associated with higher gradients across the aortic valve. Metabolomic profiles revealed that increased glycoprotein acetyls were also associated with an increased aortic valve mean gradient (0.92 SD, P=2.1 x 10−22). Finally, velocity-derived phenotypes were risk markers for aortic and mitral valve surgery even at thresholds below what is considered relevant disease currently.Using machine learning to quantify the rich phenotypic data of the UK Biobank, we report the largest assessment of valvular function and cardiovascular disease in the general population.

Background A parental history of cardiovascular disease (CVD) confers greater risk of future CVD among offspring. Whether the presence of parental modifiable risk factors contribute to or modify CVD risk in offspring is unclear. Methods and Results We studied 6278 parent–child trios in the multigenerational longitudinal Framingham Heart Study. We assessed parental history of CVD and modifiable risk factors (smoking, hypertension, diabetes, obesity, and hyperlipidemia). Multivariable Cox models were used to evaluate the association of parental history and future CVD among offspring. Among 6278 individuals (mean age 45±11 years), 44% had at least 1 parent with history of CVD. Over a median follow‐up of 15 years, 353 major CVD events occurred among offspring. Parental history of CVD conferred 1.7‐fold increased hazard of future CVD (hazard ratio [HR], 1.71 [95% CI, 1.33–2.21]). Parental obesity and smoking status were associated with higher hazard of future CVD (obesity: HR, 1.32 [95% CI, 1.06–1.64]; smoking: HR, 1.34 [95% CI, 1.07–1.68], attenuated after adjusting for offspring smoking status). By contrast, parental history of hypertension, diabetes, and hypercholesterolemia were not associated with future CVD in offspring ( P >0.05 for all). Furthermore, parental risk factors did not modify the association of parental CVD history on future offspring CVD risk. Conclusions Parental history of obesity and smoking were associated with a higher hazard of future CVD in offspring. By contrast, other parental modifiable risk factors did not alter offspring CVD risk. In addition to parental CVD, the presence of parental obesity should prompt a focus on disease prevention.

Increased left ventricular (LV) mass (LVM) and LV hypertrophy (LVH) are risk markers for adverse cardiovascular events, and may indicate an underlying cardiomyopathy. Cardiac magnetic resonance (CMR) is the gold standard for LVM estimation, but is challenging to obtain at scale, which has limited the power of prior genetic analyses. In the current study, we performed a genome-wide association study (GWAS) of CMR-derived LVM indexed to body surface area (LVMI) estimated using a deep learning algorithm within nearly 50,000 participants from the UK Biobank. We identified 12 independent associations (1 known atTTNand 11 novel) meeting genome-wide significance, implicating several candidate genes previously associated with cardiac contractility and cardiomyopathy. Greater CMR-derived LVMI was associated with higher risk of incident dilated (hazard ratio [HR] 2.58 per 1-SD increase, 95% CI 2.10-3.17) and hypertrophic (HR 2.62, 95% CI 2.09-3.30) cardiomyopathies. A polygenic risk score (PRS) for LVMI was also associated with incident hypertrophic cardiomyopathy within a separate set of UK Biobank participants (HR 1.12, 95% CI 1.01-1.12) and among individuals in an external Mass General Brigham dataset (HR 1.18, 95% CI 1.01-1.37). In summary, using CMR-derived LVM available at scale, we have identified 12 common variants associated with LVMI (11 novel) and demonstrated that both CMR-derived and genetically determined LVMI are associated with risk of incident cardiomyopathy.Journal Subject Termsmachine learning, left ventricular hypertrophy, genetics

BACKGROUND: Differentiating among cardiac diseases associated with left ventricular hypertrophy (LVH) informs diagnosis and clinical care. OBJECTIVE: To evaluate if artificial intelligence–enabled analysis of the 12-lead electrocardiogram (ECG) facilitates automated detection and classification of LVH. METHODS: We used a pretrained convolutional neural network to derive numerical representations of 12-lead ECG waveforms from patients in a multi-institutional healthcare system who had cardiac diseases associated with LVH (n = 50,709), including cardiac amyloidosis (n = 304), hypertrophic cardiomyopathy (n = 1056), hypertension (n = 20,802), aortic stenosis (n = 446), and other causes (n = 4766). We then regressed LVH etiologies relative to no LVH on age, sex, and the numerical 12-lead representations using logistic regression (“LVH-Net”). To assess deep learning model performance on single-lead data analogous to mobile ECGs, we also developed 2 single-lead deep learning models by training models on lead I (“LVH-Net Lead I”) or lead II (“LVH-Net Lead II”) from the 12-lead ECG. We compared the performance of the LVH-Net models to alternative models fit on (1) age, sex, and standard ECG measures, and (2) clinical ECG-based rules for diagnosing LVH. RESULTS: The areas under the receiver operator characteristic curve of LVH-Net by specific LVH etiology were cardiac amyloidosis 0.95 [95% CI, 0.93–0.97], hypertrophic cardiomyopathy 0.92 [95% CI, 0.90–0.94], aortic stenosis LVH 0.90 [95% CI, 0.88-0.92], hypertensive LVH 0.76 [95% CI, 0.76-0.77], and other LVH 0.69 [95% CI 0.68-0.71]. The single-lead models also discriminated LVH etiologies well. CONCLUSION: An artificial intelligence–enabled ECG model is favorable for detection and classification of LVH and outperforms clinical ECG-based rules.

This study aimed to evaluate hemodynamic correlates of inducible blood pressure (BP) pulsatility with exercise in heart failure with preserved ejection fraction (HFpEF), to identify relationships to outcomes, and to compare this with heart failure with reduced ejection fraction (HFrEF).In HFpEF, determinants and consequences of exercise BP pulsatility are not well understood.We measured exercise BP in 146 patients with HFpEF who underwent invasive cardiopulmonary exercise testing. Pulsatile BP was evaluated as proportionate pulse pressure (PrPP), the ratio of pulse pressure to systolic pressure. We measured pulmonary arterial catheter pressures, Fick cardiac output, respiratory gas exchange, and arterial stiffness. We correlated BP changes to central hemodynamics and cardiovascular outcome (nonelective cardiovascular hospitalization) and compared findings with 57 patients with HFrEF from the same referral population.In HFpEF, only age (standardized beta = 0.593; P 0.001), exercise stroke volume (standardized beta = 0.349; P 0.001), and baseline arterial stiffness (standardized beta = 0.182; P = 0.02) were significant predictors of peak exercise PrPP in multivariable analysis (R = 0.661). In HFpEF, lower PrPP was associated with lower risk of cardiovascular events, despite adjustment for confounders (HR:0.53 for PrPP below median; 95% CI: 0.28-0.98; P = 0.043). In HFrEF, lower exercise PrPP was not associated with arterial stiffness but was associated with lower peak exercise stroke volume (P = 0.013) and higher risk of adverse cardiovascular outcomes (P = 0.004).In HFpEF, greater inducible BP pulsatility measured using exercise PrPP reflects greater arterial stiffness and higher risk of adverse cardiovascular outcomes, in contrast to HFrEF where inducible exercise BP pulsatility relates to stroke volume reserve and favorable outcome.

There is growing recognition that reproductive factors are associated with increased risk of future cardiovascular disease. Infertility has been less well studied, although emerging data support its association with increased risk of cardiovascular disease. Whether infertility is associated with future risk of heart failure (HF) is not known.This study sought to examine the development of HF and HF subtypes in women with and without history of infertility.We followed postmenopausal women from the Women's Health Initiative prospectively for the development of HF. Infertility was self-reported at study baseline. Multivariable cause-specific Cox models were used to evaluate the association of infertility with incident overall HF and HF subtypes (heart failure with preserved ejection fraction [HFpEF]: left ventricular ejection fraction of ≥50% vs heart failure with reduced ejection fraction [HFrEF]: left ventricular ejection fraction of 50%]).Among 38,528 postmenopausal women (mean age: 63 ± 7 years), 5,399 (14%) participants reported a history of infertility. Over a median follow-up of 15 years, 2,373 developed incident HF, including 807 with HFrEF and 1,133 with HFpEF. Infertility was independently associated with future risk of overall HF (HR: 1.16; 95% CI: 1.04-1.30; P = 0.006). Notably, when examining HF subtypes, infertility was associated with future risk of HFpEF (HR: 1.27; 95% CI: 1.09-1.48; P = 0.002) but not HFrEF (HR: 0.97; 95% CI: 0.80-1.18).Infertility was significantly associated with incident HF. This was driven by increased risk of HFpEF, but not HFrEF, and appeared independent of traditional cardiovascular risk factors and other infertility-related conditions. Future research should investigate mechanisms that underlie the link between infertility and HFpEF.

International migration has meant that many transnational families develop transnational circuits of care to maintain collective family welfare. Although the emotional toll of geographical separation on the family has been recognized, the perspectives of elderly family members have remained relatively under-explored. Our paper seeks to plug this gap by first studying how emotions mediate the impact of geographical distance on caregiving and conversely, how distance modulates emotions related to receiving care. Second, it examines the aging futures that the elderly envisage, including the emotions that they negotiate, particularly as they anticipate changing health situations. Our analysis draws on in-depth interviews with 17 older Singaporeans (aged from the mid-60s to mid-90s) with at least one adult child residing overseas to highlight the emotional complexities of eldercare in the context of transnational families.

Sex-based differences in cardiovascular disease presentation, diagnosis, and response to therapies are well established, but mechanistic understanding and translation to clinical applications are limited. Blood-based biomarkers have become an important tool for interrogating biologic pathways. Understanding sexual dimorphism in the relationship between biomarkers and cardiovascular disease will enhance our insights into cardiovascular disease pathogenesis in women, with potential to translate to improved individualized care for men and women with or at risk for cardiovascular disease. In this review, we examine how biologic sex associates with differential levels of blood-based biomarkers and influences the effect of biomarkers on disease outcomes. We further summarize key differences in blood-based cardiovascular biomarkers along central biologic pathways, including myocardial stretch/injury, inflammation, adipose tissue metabolism, and fibrosis pathways in men versus women. Finally, we present recommendations for leveraging our current knowledge of sex differences in blood-based biomarkers for future research and clinical innovation.

On the basis of an extensive academic–public health partnership around COVID-19 response, we illustrate the challenge of science-policy translation by examining one of the most common nonpharmaceutical interventions: capacity limits. We study the implementation of a 20% capacity limit in retail facilities in the California Bay Area. Through a difference-in-differences analysis, we show that the intervention caused no material reduction in visits, using the same large-scale mobile device data on human movements (mobility data) originally used in the academic literature to support such limits. We show that the lack of effectiveness stems from a mismatch between the academic metric of capacity relative to peak visits and the policy metric of capacity relative to building code. The disconnect in metrics is amplified by mobility data losing predictive power after the early months of the pandemic, weakening the policy relevance of mobility-based interventions. Nonetheless, the data suggest that a better-grounded rationale for capacity limits is to reduce risk specifically during peak hours. To enhance the connection between science, policy, and public health in future times of crisis, we spell out 3 strategies: living models, coproduction, and shared metrics. (Am J Public Health. 2022;112(2):308–315. https://doi.org/10.2105/AJPH.2021.306576 )

Mechanisms underlying sex differences in heart failure with preserved ejection fraction (HFpEF) are poorly understood. We sought to examine sex differences in measures of arterial stiffness and the association of arterial stiffness measures with left ventricular hemodynamic responses to exercise in men and women.We studied 83 men (mean age 62 years) and 107 women (mean age 59 years) with HFpEF who underwent cardiopulmonary exercise testing with invasive hemodynamic monitoring and arterial stiffness measurement (augmentation pressure [AP], augmentation index [AIx], and aortic pulse pressure [AoPP]). Sex differences were compared using multivariable linear regression. We examined the association of arterial stiffness with abnormal left ventricular diastolic response to exercise, defined as a rise in pulmonary capillary wedge pressure relative to cardiac output (∆PCWP/∆CO) ≥ 2 mmHg/L/min by using logistic regression models.Women with HFpEF had increased arterial stiffness compared with men. AP was nearly 10 mmHg higher, and AIx was more than 10% higher in women compared with men (P0.0001 for both). Arterial stiffness measures were associated with a greater pulmonary capillary wedge pressure response to exercise, particularly among women. A 1-standard deviation higher AP was associated with3-fold increased odds of abnormal diastolic exercise response (AP: OR 3.16, 95% CI 1.34-7.42; P = 0.008 [women] vs OR 2.07, 95% CI 0.95-5.49; P = 0.15 [men]) with similar findings for AIx and AoPP.Arterial stiffness measures are significantly higher in women with HFpEF than in men and are associated with abnormally steep increases in pulmonary capillary wedge pressure with exercise, particularly in women. Arterial stiffness may preferentially contribute to abnormal diastolic function during exercise in women with HFpEF compared with men.

IntroductionRight ventricular (RV) dysfunction is associated with increased mortality across a spectrum of cardiovascular diseases. The role of obesity in RV dysfunction and adverse outcomes is unclear.MethodsWe examined patients undergoing right heart catheterization between 2005-2016 in a hospital-based cohort. Linear regression was used to examine the association of obesity with hemodynamic indices of RV dysfunction [pulmonary artery pulsatility index (PAPi), right atrial pressure: pulmonary capillary wedge pressure ratio (RAP:PCWP), RV stroke work index (RVSWI)]. Cox models were used to examine the association of RV function measures with clinical outcomes.ResultsAmong 8285 patients (mean age 63 years, 40% women), higher BMI was associated with worse indices of RV dysfunction, including lower PAPi (β -0.26, SE 0.01, p ConclusionsPatients with obesity had worse hemodynamic measured indices of RV function across a broad hospital-based sample. While RV dysfunction was associated with worse clinical outcomes including mortality and HF hospitalization, this association was especially pronounced among individuals with higher BMI.

Background: Artificial intelligence (AI)–enabled analysis of 12-lead ECGs may facilitate efficient estimation of incident atrial fibrillation (AF) risk. However, it remains unclear whether AI provides meaningful and generalizable improvement in predictive accuracy beyond clinical risk factors for AF. Methods: We trained a convolutional neural network (ECG-AI) to infer 5-year incident AF risk using 12-lead ECGs in patients receiving longitudinal primary care at Massachusetts General Hospital (MGH). We then fit 3 Cox proportional hazards models, composed of ECG-AI 5-year AF probability, CHARGE-AF clinical risk score (Cohorts for Heart and Aging in Genomic Epidemiology–Atrial Fibrillation), and terms for both ECG-AI and CHARGE-AF (CH-AI), respectively. We assessed model performance by calculating discrimination (area under the receiver operating characteristic curve) and calibration in an internal test set and 2 external test sets (Brigham and Women’s Hospital [BWH] and UK Biobank). Models were recalibrated to estimate 2-year AF risk in the UK Biobank given limited available follow-up. We used saliency mapping to identify ECG features most influential on ECG-AI risk predictions and assessed correlation between ECG-AI and CHARGE-AF linear predictors. Results: The training set comprised 45 770 individuals (age 55±17 years, 53% women, 2171 AF events) and the test sets comprised 83 162 individuals (age 59±13 years, 56% women, 2424 AF events). Area under the receiver operating characteristic curve was comparable using CHARGE-AF (MGH, 0.802 [95% CI, 0.767–0.836]; BWH, 0.752 [95% CI, 0.741–0.763]; UK Biobank, 0.732 [95% CI, 0.704–0.759]) and ECG-AI (MGH, 0.823 [95% CI, 0.790–0.856]; BWH, 0.747 [95% CI, 0.736–0.759]; UK Biobank, 0.705 [95% CI, 0.673–0.737]). Area under the receiver operating characteristic curve was highest using CH-AI (MGH, 0.838 [95% CI, 0.807 to 0.869]; BWH, 0.777 [95% CI, 0.766 to 0.788]; UK Biobank, 0.746 [95% CI, 0.716 to 0.776]). Calibration error was low using ECG-AI (MGH, 0.0212; BWH, 0.0129; UK Biobank, 0.0035) and CH-AI (MGH, 0.012; BWH, 0.0108; UK Biobank, 0.0001). In saliency analyses, the ECG P-wave had the greatest influence on AI model predictions. ECG-AI and CHARGE-AF linear predictors were correlated (Pearson r : MGH, 0.61; BWH, 0.66; UK Biobank, 0.41). Conclusions: AI-based analysis of 12-lead ECGs has similar predictive usefulness to a clinical risk factor model for incident AF and the approaches are complementary. ECG-AI may enable efficient quantification of future AF risk.