Scalable Risk Stratification for Heart Failure Using Artificial Intelligence applied to 12-lead Electrocardiographic Images: A Multinational Study.

Background: Current risk stratification strategies for heart failure (HF) risk require either specific blood-based biomarkers or comprehensive clinical evaluation. In this study, we evaluated the use of artificial intelligence (AI) applied to images of electrocardiograms (ECGs) to predict HF risk. Methods: Across multinational longitudinal cohorts in the integrated Yale New Haven Health System (YNHHS) and in population-based UK Biobank (UKB) and Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), we identified individuals without HF at baseline. Incident HF was defined based on the first occurrence of an HF hospitalization. We evaluated an AI-ECG model that defines the cross-sectional probability of left ventricular dysfunction from a single image of a 12-lead ECG and its association with incident HF. We accounted for the competing risk of death using the Fine-Gray subdistribution model and evaluated the discrimination using Harrel's c-statistic. The pooled cohort equations to prevent HF (PCP-HF) were used as a comparator for estimating incident HF risk. Results: Among 231,285 individuals at YNHHS, 4472 had a primary HF hospitalization over 4.5 years (IQR 2.5-6.6) of follow-up. In UKB and ELSA-Brasil, among 42,741 and 13,454 people, 46 and 31 developed HF over a follow-up of 3.1 (2.1-4.5) and 4.2 (3.7-4.5) years, respectively. A positive AI-ECG screen portended a 4-fold higher risk of incident HF among YNHHS patients (age-, sex-adjusted HR [aHR] 3.88 [95% CI, 3.63-4.14]). In UKB and ELSA-Brasil, a positive-screen ECG portended 13- and 24-fold higher hazard of incident HF, respectively (aHR: UKBB, 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 model output probabilities were progressively associated with a higher risk for HF. The model's discrimination for incident HF was 0.718 in YNHHS, 0.769 in UKB, and 0.810 in ELSA-Brasil. Across cohorts, incorporating model probability with PCP-HF yielded a significant improvement in discrimination over PCP-HF alone. Conclusions: An AI model applied to images of 12-lead ECGs can identify those at elevated risk of HF across multinational cohorts. As a digital biomarker of HF risk that requires just an ECG image, this AI-ECG approach can enable scalable and efficient screening for HF risk.

Combination of Tele-Cardiology Tools for Cardiovascular Risk Stratification in Primary Care: Data from the PROVAR+ Study. / Combinação de Ferramentas de Telecardiologia para Estratificação de Risco Cardiovascular na Atenção Primária: Dados do Estudo PROVAR.

BACKGROUND: Tele-cardiology tools are valuable strategies to improve risk stratification. OBJECTIVE: We aimed to evaluate the accuracy of tele-electrocardiography (ECG) to predict abnormalities in screening echocardiography (echo) in primary care (PC). METHODS: In 17 months, 6 health providers at 16 PC units were trained on simplified handheld echo protocols. Tele-ECGs were recorded for final diagnosis by a cardiologist. Consented patients with major ECG abnormalities by the Minnesota code, and a 1:5 sample of normal individuals underwent clinical questionnaire and screening echo interpreted remotely. Major heart disease was defined as moderate/severe valve disease, ventricular dysfunction/hypertrophy, pericardial effusion, or wall-motion abnormalities. Association between major ECG and echo abnormalities was assessed by logistic regression as follows: 1) unadjusted model; 2) model 1 adjusted for age/sex; 3) model 2 plus risk factors (hypertension/diabetes); 4) model 3 plus history of cardiovascular disease (Chagas/rheumatic heart disease/ischemic heart disease/stroke/heart failure). P-values < 0.05 were considered significant. RESULTS: A total 1,411 patients underwent echo; 1,149 (81%) had major ECG abnormalities. Median age was 67 (IQR 60 to 74) years, and 51.4% were male. Major ECG abnormalities were associated with a 2.4-fold chance of major heart disease on echo in bivariate analysis (OR = 2.42 [95% CI 1.76 to 3.39]), and remained significant after adjustments in models (p < 0.001) 2 (OR = 2.57 [95% CI 1.84 to 3.65]), model 3 (OR = 2.52 [95% CI 1.80 to3.58]), and model 4 (OR = 2.23 [95%CI 1.59 to 3.19]). Age, male sex, heart failure, and ischemic heart disease were also independent predictors of major heart disease on echo. CONCLUSIONS: Tele-ECG abnormalities increased the likelihood of major heart disease on screening echo, even after adjustments for demographic and clinical variables.

FUNDAMENTO: As ferramentas de telecardiologia são estratégias valiosas para melhorar a estratificação de risco. OBJETIVO: Objetivamos avaliar a acurácia da tele-eletrocardiografia (ECG) para predizer anormalidades no ecocardiograma de rastreamento na atenção primária. MÉTODOS: Em 17 meses, 6 profissionais de saúde em 16 unidades de atenção primária foram treinados em protocolos simplificados de ecocardiografia portátil. Tele-ECGs foram registrados para diagnóstico final por um cardiologista. Pacientes consentidos com anormalidades maiores no ECG pelo código de Minnesota e uma amostra 1:5 de indivíduos normais foram submetidos a um questionário clínico e ecocardiograma de rastreamento interpretado remotamente. A doença cardíaca grave foi definida como doença valvular moderada/grave, disfunção/hipertrofia ventricular, derrame pericárdico ou anormalidade da motilidade. A associação entre alterações maiores do ECG e anormalidades ecocardiográficas foi avaliada por regressão logística da seguinte forma: 1) modelo não ajustado; 2) modelo 1 ajustado por idade/sexo; 3) modelo 2 mais fatores de risco (hipertensão/diabetes); 4) modelo 3 mais história de doença cardiovascular (Chagas/cardiopatia reumática/cardiopatia isquêmica/AVC/insuficiência cardíaca). Foram considerados significativos valores de p < 0,05. RESULTADOS: No total, 1.411 pacientes realizaram ecocardiograma, sendo 1.149 (81%) com anormalidades maiores no ECG. A idade mediana foi de 67 anos (intervalo interquartil de 60 a 74) e 51,4% eram do sexo masculino. As anormalidades maiores no ECG se associaram a uma chance 2,4 vezes maior de doença cardíaca grave no ecocardiograma de rastreamento na análise bivariada (OR = 2,42 [IC 95% 1,76 a 3,39]) e permaneceram significativas (p < 0,001) após ajustes no modelo 2 (OR = 2,57 [IC 95% 1,84 a 3,65]), modelo 3 (OR = 2,52 [IC 95% 1,80 a 3,58]) e modelo 4 (OR = 2,23 [IC 95% 1,59 a 3,19]). Idade, sexo masculino, insuficiência cardíaca e doença cardíaca isquêmica também foram preditores independentes de doença cardíaca grave no ecocardiograma. CONCLUSÕES: As anormalidades do tele-ECG aumentaram a probabilidade de doença cardíaca grave no ecocardiograma de rastreamento, mesmo após ajustes para variáveis demográficas e clínicas.

Evaluation of patients with implantable cardioverter-defibrillator in a Latin American tertiary center.

INTRODUCTION: Despite advancements in implantable cardioverter-defibrillator (ICD) technology, sudden cardiac death (SCD) remains a persistent public health concern. Chagas disease (ChD), prevalent in Brazil, is associated with increased ventricular tachycardia (VT) and ventricular fibrillation (VF) events and SCD compared to other cardiomyopathies. METHODS: This retrospective observational study included patients who received ICDs between October 2007 and December 2018. The study aims to assess whether mortality and VT/VF events decreased in patients who received ICDs during different time periods (2007-2010, 2011-2014, and 2015-2018). Additionally, it seeks to compare the prognosis of ChD patients with non-ChD patients. Time periods were chosen based on the establishment of the Arrhythmia Service in 2011. The primary outcome was overall mortality, assessed across the entire sample and the three periods. Secondary outcomes included VT/VF events and the combined outcome of death or VT/VF. RESULTS: Of the 885 patients included, 31% had ChD. Among them, 28% died, 14% had VT/VF events, and 37% experienced death and/or VT/VF. Analysis revealed that period 3 (2015-2018) was associated with better death-free survival (p = .007). ChD was the only variable associated with a higher rate of VT/VF events (p < .001) and the combined outcome (p = .009). CONCLUSION: Mortality and combined outcome rates decreased gradually for ICD patients during the periods 2011-2014 and 2015-2018 compared to the initial period (2007-2010). ChD was associated with higher VT/VF events in ICD patients, only in the first two periods.

Agreement between Handheld and Standard Echocardiography for Diagnosis of Latent Rheumatic Heart Disease in Brazilian Schoolchildren from High-Prevalence Settings (Agreement between Screening and Standard Echo for RHD).

Introduction: Handheld echocardiography (echo) is the tool of choice for rheumatic heart disease (RHD) screening. We aimed to assess the agreement between screening and standard echo for latent RHD diagnosis in schoolchildren from an endemic setting. Methods: Over 14 months, 3 nonphysicians used handheld machines and the 2012 WHF Criteria to determine RHD prevalence in consented schoolchildren from Brazilian low-income public schools. Studies were interpreted by telemedicine by 3 experts (Brazil, US). RHD-positive children (borderline/definite) and those with congenital heart disease (CHD) were referred for standard echo, acquired and interpreted by a cardiologist. Agreement between screening and standard echo, by WHF subgroups, was assessed. Results: 1390 students were screened in 6 schools, with 110 (7.9%, 95% CI 6.5-9.5) being screen positive (14 ± 2 years, 72% women). Among 16 cases initially diagnosed as definite RHD, 11 (69%) were confirmed, 4 (25%) reclassified to borderline, and 1 to normal. Among 79 cases flagged as borderline RHD, 19 (24%) were confirmed, 50 (63%) reclassified to normal, 8 (10%) reclassified as definite RHD, and 2 had mild CHD. Considering the 4 diagnostic categories, kappa was 0.18. In patients with borderline RHD reclassified to non-RHD, the most frequent WHF criterion was B (isolated mitral regurgitation, 64%), followed by A (2 mitral valve morphological features, 31%). In 1 patient with definite RHD reclassified to normal, the WHF criterion was D (borderline RHD in aortic and mitral valves). After standard echo, RHD prevalence was 3.2% (95% CI 2.3-4.2). Conclusions: Although practical, RHD screening with handheld devices tends to overestimate prevalence.

Efficacy of three benznidazole dosing strategies for adults living with chronic Chagas disease (MULTIBENZ): an international, randomised, double-blind, phase 2b trial.

BACKGROUND: Treatment with benznidazole for chronic Chagas disease is associated with low cure rates and substantial toxicity. We aimed to compare the parasitological efficacy and safety of 3 different benznidazole regimens in adult patients with chronic Chagas disease. METHODS: The MULTIBENZ trial was an international, randomised, double-blind, phase 2b trial performed in Argentina, Brazil, Colombia, and Spain. We included participants aged 18 years and older diagnosed with Chagas disease with two different serological tests and detectable T cruzi DNA by qPCR in blood. Previously treated people, pregnant women, and people with severe cardiac forms were excluded. Participants were randomly assigned 1:1:1, using a balanced block randomisation scheme stratified by country, to receive benznidazole at three different doses: 300 mg/day for 60 days (control group), 150 mg/day for 60 days (low dose group), or 400 mg/day for 15 days (short treatment group). The primary outcome was the proportion of patients with a sustained parasitological negativity by qPCR during a follow-up period of 12 months. The primary safety outcome was the proportion of people who permanently discontinued the treatment. Both primary efficacy analysis and primary safety analysis were done in the intention-to-treat population. The trial is registered with EudraCT, 2016-003789-21, and ClinicalTrials.gov, NCT03191162, and is completed. FINDINGS: From April 20, 2017, to Sept 20, 2020, 245 people were enrolled, and 234 were randomly assigned: 78 to the control group, 77 to the low dose group, and 79 to the short treatment group. Sustained parasitological negativity was observed in 42 (54%) of 78 participants in the control group, 47 (61%) of 77 in the low dose group, and 46 (58%) of 79 in the short treatment group. Odds ratios were 1·41 (95% CI 0·69-2·88; p=0·34) when comparing the low dose and control groups and 1·23 (0·61-2·50; p=0·55) when comparing short treatment and control groups. 177 participants (76%) had an adverse event: 62 (79%) in the control group, 56 (73%) in the low dose group, and 59 (77%) in the short treatment group. However, discontinuations were less frequent in the short treatment group compared with the control group (2 [2%] vs 11 [14%]; OR 0·20, 95% CI 0·04-0·95; p=0·044). INTERPRETATION: Participants had a similar parasitological responses. However, reducing the usual treatment from 8 weeks to 2 weeks might maintain the same response while facilitating adherence and increasing treatment coverage. These findings should be confirmed in a phase 3 clinical trial. FUNDING: European Community's 7th Framework Programme.

Gastrointestinal Manifestations of Chagas Disease: A Systematic Review with Meta-Analysis.

The aims of this study were to estimate the prevalence of gastrointestinal manifestations among individuals with positive serology for Chagas disease (ChD) and to describe the clinical gastrointestinal manifestations of the disease. A systematic review with meta-analysis was conducted based on the criteria and recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines. The PubMed, Scopus, Virtual Health Library, Web of Science, and Embase databases were used to search for evidence. Two reviewers independently selected eligible articles and extracted data. RStudio® software was used for the meta-analysis. For subgroup analysis, the studies were divided according to the origin of the individuals included: 1) individuals from health units were included in the health care service prevalence analysis, and 2) individuals from the general population were included in the population prevalence analysis. A total of 2,570 articles were identified, but after removal of duplicates and application of inclusion criteria, 24 articles were included and 21 were part of the meta-analysis. Most of the studies were conducted in Brazil. Radiological diagnosis was the most frequent method used to identify the gastrointestinal clinical form. The combined effect of meta-analysis studies showed a prevalence of gastrointestinal manifestations in individuals with ChD of 12% (95% CI, 8.0-17.0%). In subgroup analysis, the prevalence for studies involving health care services was 16% (95% CI, 11.0-23.0%), while the prevalence for population-based studies was 9% (95% CI, 5.0-15.0%). Megaesophagus and megacolon were the main forms of ChD presentation in the gastrointestinal form. The prevalence of gastrointestinal manifestations of ChD was 12%. Knowing the prevalence of ChD in its gastrointestinal form is an important step in planning health actions for these patients.

Vagal dysautonomia in patients with Chagas disease and mortality: 14-year results of a population cohort of the elderly.

INTRODUCTION: Great part of Chagas disease (ChD) mortality occurs due to ventricular arrhythmias, and autonomic function (AF) may predict unfavorable outcomes. We aimed to evaluate the predictive value of AF indexes in ChD patients. METHODS: The Bambuí Study of Aging is a prospective cohort of residents ≥60 years at study onset (1997), in the southeastern Brazilian city of Bambuí (15,000 inhabitants). Consented participants underwent annual follow-up visits, and death certificates were tracked. AF was assessed by the maximum expiration on minimum inspiration (E:I) ratio during ECG acquisition and by heart rate variability indices: SDRR (standard deviation of adjacent RR intervals) and RMSSD (square root of the mean of the sum of squares of the differences between adjacent RR intervals)), calculated using a computer algorithm. Cox proportional hazards regression was performed to access the prognostic value of AF indexes, expressed as terciles, for all-cause mortality, after adjustment for demographic, clinical and ECG variables. RESULTS: From 1742 qualifying residents, 1000 had valid AF tests, being 321 with ChD. Among these, median age was 68 (64-74) years, and 32.5% were men. In Cox survival analyses, only SDRR was associated with all-cause mortality in non-adjusted models: SDRR (hazard ratio (HR): 1.26 (95% CI 1.08-1.47), p < 0.001), E:I ratio (HR: 1.13 (95% CI 0,98-1.31), p = 0.10) and RMSSD (HR: 0.99 (0.86-1.16), p = 0.95). After adjustment for sex and age, none of the indexes remained as independent predictors. CONCLUSION: Among elderly patients with ChD, AF indexes available in this cohort were not independent predictors of 14-year mortality.

Adherence to the Dietary Approaches to Stop Hypertension (DASH) and Serum Urate Concentrations: A Longitudinal Analysis from the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil).

BACKGROUND: Increased serum urate (SU) and hyperuricemia (HU) are associated with chronic noncommunicable diseases and mortality. SU concentrations are affected by several factors, including diet, and are expected to rise with age. We investigated whether the Dietary Approaches to Stop Hypertension (DASH) diet alter this trend. OBJECTIVE: The objective was to assess whether adherence to the DASH diet predicts a longitudinal change in SU concentrations and risk of HU in 8 y of follow-up. METHODS: Longitudinal analyses using baseline (2008-2010, aged 35-74 y), second (2012-2014), and third (2016-2018) visits data from the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). The inclusion criteria were having complete food frequency questionnaire (FFQ) and urinary sodium measurement, in addition to having SU measurement at the 1st visit and at least 1 of the 2 follow-up visits. For the HU incidence analyses, participants had also to be free from HU at baseline. The final samples included 12575 individuals for the SU change analyses and 10549 for the HU incidence analyses. Adherence to DASH diet was assessed as continuous value. HU was defined as SU>6.8 mg/dL and/or urate-lowering therapy use. Mixed-effect linear and Poisson regressions (incidence rate ratio [IRR] and 95% confidence interval [CI]) were used in the analyses, adjusted for confounders. RESULTS: The mean age was 51.4 (8.7) y, and 55.4% were females. SU means (standard deviation) were 5.4 (1.4) at 1st visit, 5.2 (1.4) at 2nd visit, and 5.1(1.3) mg/dL at 3rd visit. The HU incidence rate was 8.87 per 1000 person-y. Each additional point in adherence to the DASH diet accelerated SU decline (P< 0.01) and lowered the incidence of HU by 4.3% (IRR: 0.957; 95% CI: 0.938,0.977) in adjusted model. CONCLUSION: The present study findings reinforce the importance of encouraging the DASH diet as a healthy dietary pattern to control and reduce the SU concentrations and risk of HU.

Outcomes of latent rheumatic heart disease: External validation of a simplified score in patients with and without secondary prophylaxis.

BACKGROUND: Secondary antibiotic prophylaxis reduces progression of latent rheumatic heart disease (RHD) but not all children benefit. Improved risk stratification could refine recommendations following positive screening. We aimed to evaluate the performance of a previously developed echocardiographic risk score to predict mid-term outcomes among children with latent RHD. METHODS: We included children who completed the GOAL, a randomized trial of secondary antibiotic prophylaxis among children with latent RHD in Uganda. Outcomes were determined by a 4-member adjudication panel. We applied the point-based score, consisting of 5 variables (mitral valve (MV) anterior leaflet thickening (3 points), MV excessive leaflet tip motion (3 points), MV regurgitation jet length ≥ 2 cm (6 points), aortic valve focal thickening (4 points) and any aortic regurgitation (5 points)), to panel results. Unfavorable outcome was defined as progression of diagnostic category (borderline to definite, mild definite to moderate/severe definite), worsening valve involvement or remaining with mild definite RHD. RESULTS: 799 patients (625 borderline and 174 definite RHD) were included, with median follow-up of 24 months. At total 116 patients (14.5%) had unfavorable outcome per study criteria, 57.8% not under prophylaxis. The score was strongly associated with unfavorable outcome (HR = 1.26, 95% CI 1.16-1.37, p < 0.001). Unfavorable outcome rates in low (≤6 points), intermediate (7-9 points) and high-risk (≥10 points) children at follow-up were 11.8%, 30.4%, and 42.2%, (p < 0.001) respectively (C-statistic = 0.64 (95% CI 0.59-0.69)). CONCLUSIONS: The simple risk score provided an accurate prediction of RHD status at 2-years, showing a good performance in a population with milder RHD phenotypes.

Percentiles of predicted 10-year cardiovascular disease risk by sex and age in Brazil and their association with estimated risk of long-term atherosclerotic events.

OBJECTIVE: Expressing the cardiovascular disease (CVD) risk in relation to peers may complement the estimation of absolute CVD risk. We aimed to determine 10-year CVD risk percentiles by sex and age in the Brazilian population and evaluate their association with estimated long-term atherosclerotic CVD (ASCVD) risk. METHODS: A cross-sectional analysis of baseline data from the ELSA-Brasil study was conducted in individuals aged 40-74 years without prior ASCVD. Ten-year CVD risk and long-term ASCVD risk were estimated by the WHO risk score and the Multinational Cardiovascular Risk Consortium tool, respectively. Ten-year risk percentiles were determined by ranking the calculated risks within each sex and age group. RESULTS: Ten-year CVD risk versus percentile plots were constructed for each sex and age group using data from 13,364 participants (55% females; median age, 52 [IQR, 46-59] years). Long-term ASCVD risk was calculated in 12,973 (97.1%) participants. Compared to individuals at the <25th risk percentile, those at the ≥75th percentile had a greater risk of being in the highest quartile of long-term risk (ORs [95% CIs] 6.57 [5.18-8.30] in females and 11.59 [8.42-15.96] in males) in regression models adjusted for age, race, education, and 10-year CVD risk. In both sexes, the association between risk percentile and long-term risk weakened after age 50. A tool for calculating 10-year CVD risk and the corresponding percentile is available at https://bit.ly/3CzPUi6. CONCLUSIONS: We established percentiles of predicted 10-year CVD risk by sex and age in the Brazilian population, which independently reflect the estimated long-term ASCVD risk in younger individuals.

Heart age gap estimated by explainable advanced electrocardiography is associated with cardiovascular risk factors and survival.

Aims: Deep neural network artificial intelligence (DNN-AI)-based Heart Age estimations have been presented and used to show that the difference between an electrocardiogram (ECG)-estimated Heart Age and chronological age is associated with prognosis. An accurate ECG Heart Age, without DNNs, has been developed using explainable advanced ECG (A-ECG) methods. We aimed to evaluate the prognostic value of the explainable A-ECG Heart Age and compare its performance to a DNN-AI Heart Age. Methods and results: Both A-ECG and DNN-AI Heart Age were applied to patients who had undergone clinical cardiovascular magnetic resonance imaging. The association between A-ECG or DNN-AI Heart Age Gap and cardiovascular risk factors was evaluated using logistic regression. The association between Heart Age Gaps and death or heart failure (HF) hospitalization was evaluated using Cox regression adjusted for clinical covariates/comorbidities. Among patients [n = 731, 103 (14.1%) deaths, 52 (7.1%) HF hospitalizations, median (interquartile range) follow-up 5.7 (4.7-6.7) years], A-ECG Heart Age Gap was associated with risk factors and outcomes [unadjusted hazard ratio (HR) (95% confidence interval) (5 year increments): 1.23 (1.13-1.34) and adjusted HR 1.11 (1.01-1.22)]. DNN-AI Heart Age Gap was associated with risk factors and outcomes after adjustments [HR (5 year increments): 1.11 (1.01-1.21)], but not in unadjusted analyses [HR 1.00 (0.93-1.08)], making it less easily applicable in clinical practice. Conclusion: A-ECG Heart Age Gap is associated with cardiovascular risk factors and HF hospitalization or death. Explainable A-ECG Heart Age Gap has the potential for improving clinical adoption and prognostic performance compared with existing DNN-AI-type methods.

End-to-end risk prediction of atrial fibrillation from the 12-Lead ECG by deep neural networks.

BACKGROUND: Atrial fibrillation (AF) is one of the most common cardiac arrhythmias that affects millions of people each year worldwide and it is closely linked to increased risk of cardiovas- cular diseases such as stroke and heart failure. Machine learning methods have shown promising results in evaluating the risk of developing atrial fibrillation from the electrocardiogram. We aim to develop and evaluate one such algorithm on a large CODE dataset collected in Brazil. METHODS: We used the CODE cohort to develop and test a model for AF risk prediction for individual patients from the raw ECG recordings without the use of additional digital biomarkers. The cohort is a collection of ECG recordings and annotations by the Telehealth Network of Minas Gerais, in Brazil. A convolutional neural network based on a residual network architecture was implemented to produce class probabilities for the classification of AF. The probabilities were used to develop a Cox proportional hazards model and a Kaplan-Meier model to carry out survival analysis. Hence, our model is able to perform risk prediction for the development of AF in patients without the condition. RESULTS: The deep neural network model identified patients without indication of AF in the presented ECG but who will develop AF in the future with an AUC score of 0.845. From our survival model, we obtain that patients in the high-risk group (i.e. with the probability of a future AF case being >0.7) are 50% more likely to develop AF within 40 weeks, while patients belonging to the minimal-risk group (i.e. with the probability of a future AF case being less than or equal to 0.1) have >85% chance of remaining AF free up until after seven years. CONCLUSION: We developed and validated a model for AF risk prediction. If applied in clinical practice, the model possesses the potential of providing valuable and useful information in decision- making and patient management processes.

ISE/ISHNE Expert Consensus Statement on ECG Diagnosis of Left Ventricular Hypertrophy: The Change of the Paradigm. The joint paper of the International Society of Electrocardiology and the International Society for Holter Monitoring and Noninvasive Electrocardiology.

The ECG diagnosis of LVH is predominantly based on the QRS voltage criteria, i.e. the increased QRS complex amplitude in defined leads. The classical ECG diagnostic paradigm postulates that the increased left ventricular mass generates a stronger electrical field, increasing the leftward and posterior QRS forces. These increased forces are reflected in the augmented QRS amplitude in the corresponding leads. However, the clinical observations document increased QRS amplitude only in the minority of patients with LVH. The low sensitivity of voltage criteria has been repeatedly documented. We discuss possible reasons for this shortcoming and proposal of a new paradigm.

Detection of Left Ventricular Systolic Dysfunction From Electrocardiographic Images.

BACKGROUND: Left ventricular (LV) systolic dysfunction is associated with a >8-fold increased risk of heart failure and a 2-fold risk of premature death. The use of ECG signals in screening for LV systolic dysfunction is limited by their availability to clinicians. We developed a novel deep learning-based approach that can use ECG images for the screening of LV systolic dysfunction. METHODS: Using 12-lead ECGs plotted in multiple different formats, and corresponding echocardiographic data recorded within 15 days from the Yale New Haven Hospital between 2015 and 2021, we developed a convolutional neural network algorithm to detect an LV ejection fraction <40%. The model was validated within clinical settings at Yale New Haven Hospital and externally on ECG images from Cedars Sinai Medical Center in Los Angeles, CA; Lake Regional Hospital in Osage Beach, MO; Memorial Hermann Southeast Hospital in Houston, TX; and Methodist Cardiology Clinic of San Antonio, TX. In addition, it was validated in the prospective Brazilian Longitudinal Study of Adult Health. Gradient-weighted class activation mapping was used to localize class-discriminating signals on ECG images. RESULTS: Overall, 385 601 ECGs with paired echocardiograms were used for model development. The model demonstrated high discrimination across various ECG image formats and calibrations in internal validation (area under receiving operation characteristics [AUROCs], 0.91; area under precision-recall curve [AUPRC], 0.55); and external sets of ECG images from Cedars Sinai (AUROC, 0.90 and AUPRC, 0.53), outpatient Yale New Haven Hospital clinics (AUROC, 0.94 and AUPRC, 0.77), Lake Regional Hospital (AUROC, 0.90 and AUPRC, 0.88), Memorial Hermann Southeast Hospital (AUROC, 0.91 and AUPRC 0.88), Methodist Cardiology Clinic (AUROC, 0.90 and AUPRC, 0.74), and Brazilian Longitudinal Study of Adult Health cohort (AUROC, 0.95 and AUPRC, 0.45). An ECG suggestive of LV systolic dysfunction portended >27-fold higher odds of LV systolic dysfunction on transthoracic echocardiogram (odds ratio, 27.5 [95% CI, 22.3-33.9] in the held-out set). Class-discriminative patterns localized to the anterior and anteroseptal leads (V2 and V3), corresponding to the left ventricle regardless of the ECG layout. A positive ECG screen in individuals with an LV ejection fraction ≥40% at the time of initial assessment was associated with a 3.9-fold increased risk of developing incident LV systolic dysfunction in the future (hazard ratio, 3.9 [95% CI, 3.3-4.7]; median follow-up, 3.2 years). CONCLUSIONS: We developed and externally validated a deep learning model that identifies LV systolic dysfunction from ECG images. This approach represents an automated and accessible screening strategy for LV systolic dysfunction, particularly in low-resource settings.

On Merging Feature Engineering and Deep Learning for Diagnosis, Risk Prediction and Age Estimation Based on the 12-Lead ECG.

OBJECTIVE: Over the past few years, deep learning (DL) has been used extensively in research for 12-lead electrocardiogram (ECG) analysis. However, it is unclear whether the explicit or implicit claims made on DL superiority to the more classical feature engineering (FE) approaches, based on domain knowledge, hold. In addition, it remains unclear whether combining DL with FE may improve performance over a single modality. METHODS: To address these research gaps and in-line with recent major experiments, we revisited three tasks: cardiac arrhythmia diagnosis (multiclass-multilabel classification), atrial fibrillation risk prediction (binary classification), and age estimation (regression). We used an overall dataset of 2.3M 12-lead ECG recordings to train the following models for each task: i) a random forest taking FE as input; ii) an end-to-end DL model; and iii) a merged model of FE+DL. RESULTS: FE yielded comparable results to DL while necessitating significantly less data for the two classification tasks. DL outperformed FE for the regression task. For all tasks, merging FE with DL did not improve performance over DL alone. These findings were confirmed on the additional PTB-XL dataset. CONCLUSION: We found that for traditional 12-lead ECG based diagnosis tasks, DL did not yield a meaningful improvement over FE, while it improved significantly the nontraditional regression task. We also found that combining FE with DL did not improve over DL alone, which suggests that the FE was redundant with the features learned by DL. SIGNIFICANCE: Our findings provides important recommendations on 12-lead ECG based machine learning strategy and data regime to choose for a given task. When looking at maximizing performance as the end goal, if the task is nontraditional and a large dataset is available then DL is preferable. If the task is a classical one and/or a small dataset is available then a FE approach may be the better choice.

Clinical and electrocardiographic outcomes evaluated by telemedicine of outpatients with clinical suspicion of COVID-19 treated with chloroquine compounds in Brazil†.

Aims: To evaluate clinical and electrocardiographic outcomes of patients with COVID-19, comparing those using chloroquine compounds (chloroquine) to individuals without specific treatment. Methods: Outpatients with suspected COVID-19 in Brazil who had at least one tele-electrocardiography (ECG) recorded in a telehealth system were enrolled in two arms (Group 1: chloroquine and Group 2: without specific treatment) and one registry (Group 3: other treatments). Outcomes were assessed through follow-up calls (phone contact, days 3 and 14) and linkage to national mortality and hospitalization databases. The primary outcome was composed of: hospitalization, intensive care admission, mechanical ventilation, and all-cause death, and the ECG outcome was the occurrence of major abnormalities by the Minnesota code. Significant variables in univariable logistic regression were included in 4 models: 1-unadjusted; 2-adjusted for age and sex; 3-model 2 + cardiovascular risk factors and 4-model 3 + COVID-19 symptoms. Results: In 303 days, 712 (10.2%) patients were allocated in group 1, 3,623 (52.1%) in group 2 and 2,622 (37.7%) in group 3; 1,969 had successful phone follow-up (G1: 260, G2: 871, and G3: 838). A late follow-up ECG was obtained for 917 (27.2%) patients [group 1: 81 (11.4%), group 2: 512 (14.1%), group 3: 334 (12.7%)]. In adjusted models, chloroquine was independently associated with greater chance of the composite clinical outcome: phone contact (model 4): OR = 3.24 (95% CI 2.31-4.54), p < 0.001. Chloroquine was also independently associated with higher mortality, assessed by phone + administrative data (model 3): OR = 1.67 (95% CI 1.20-2.28). However, chloroquine did not associate with the occurrence of major ECG abnormalities [model 3; OR = 0.80 (95% CI 0.63-1.02, p = 0.07)]. Abstracts with partial results of this work was accepted in the American Heart Association Scientific Sessions, November 2022, in Chicago, IL, USA. Conclusion: Chloroquine was associated with a higher risk of poor outcomes in patients suspected to have COVID-19 when compared to those who received standard care. Follow-up ECGs were obtained in only 13.2% of patients and did not show any significant differences in major abnormalities amongst the three groups. In the absence of early ECG changes, other side effects, late arrhythmias or deferral of care may be hypothesized to explain the worse outcomes.

A novel algorithm to assess the quality of 12-lead ECG recordings: validation in a real telecardiology application.

Objective. Automatic detection of Electrocardiograms (ECG) quality is fundamental to minimize costs and risks related to delayed diagnosis due to low ECG quality. Most algorithms to assess ECG quality include non-intuitive parameters. Also, they were developed using data non-representative of a real-world scenario, in terms of pathological ECGs and overrepresentation of low-quality ECG. Therefore, we introduce an algorithm to assess 12-lead ECG quality, Noise Automatic Classification Algorithm (NACA) developed in Telehealth Network of Minas Gerais (TNMG).Approach. NACA estimates a signal-to-noise ratio (SNR) for each ECG lead, where 'signal' is an estimated heartbeat template, and 'noise' is the discrepancy between the template and the ECG heartbeat. Then, clinically-inspired rules based on SNR are used to classify the ECG as acceptable or unacceptable. NACA was compared with Quality Measurement Algorithm (QMA), the winner of Computing in Cardiology Challenge 2011 (ChallengeCinC) by using five metrics: sensitivity (Se), specificity (Sp), positive predictive value (PPV),F2, and cost reduction resulting from adoption of the algorithm. Two datasets were used for validation: TestTNMG, consisting of 34 310 ECGs received by TNMG (1% unacceptable and 50% pathological); ChallengeCinC, consisting of 1000 ECGs (23% unacceptable, higher than real-world scenario).Main results. Both algorithms reached a similar performance on ChallengeCinC, although NACA performed considerably better than QMA in TestTNMG (Se = 0.89 versus 0.21; Sp = 0.99 versus 0.98; PPV = 0.59 versus 0.08;F2= 0.76 versus 0.16 and cost reduction 2.3 ± 1.8% versus 0.3 ± 0.3%, respectively).Significance. Implementing of NACA in a telecardiology service results in evident health and financial benefits for the patients and the healthcare system.

What We Lost in the Fire: Endemic Tropical Heart Diseases in the Time of COVID-19.

The COVID-19 pandemic has profoundly influenced the effort to achieve global health equity. This has been particularly the case for HIV/AIDS, tuberculosis, and malaria control initiatives in low- and middle-income countries, with significant outcome setbacks seen for the first time in decades. Lost in the calls for compensatory funding increases for such programs, however, is the plight of endemic tropical heart diseases, a group of disorders that includes rheumatic heart disease, Chagas disease, and endomyocardial fibrosis. Such endemic illnesses affect millions of people around the globe and remain a source of substantial mortality, morbidity, and health disparity. Unfortunately, these conditions were already neglected before the pandemic, and thus those living with them have disproportionately suffered during the time of COVID-19. In this perspective, we briefly define endemic tropical heart diseases, summarizing their prepandemic epidemiology, funding, and control statuses. We then describe the ways in which people living with these disorders, along with the healthcare providers and researchers working to improve their outcomes, have been harmed by the ongoing COVID-19 pandemic. We conclude by proposing the path forward, including approaches we may use to leverage lessons learned from the pandemic to strengthen care systems for these neglected diseases.

Reasons for disagreement between screening and standard echocardiography in primary care: data from the PROVAR + study : Disagreement between screening and standard echo.

We aimed to evaluate the reasons for disagreement between screening echocardiography (echo), acquired by nonexperts, and standard echo in the Brazilian primary care (PC). Over 20 months, 22 PC workers were trained on simplified handheld (GE VSCAN) echo protocols. Screening groups, consisting of patients aged 17-20, 35-40 and 60-65 years, and patients referred for clinical indications underwent focused echo. Studies were remotelyinterpreted in US and Brazil, and those diagnosed with major or severe HD were referred for standard echoperformed by an expert. Major HD was defined as moderate to severe valve disease, ventriculardysfunction/hypertrophy, pericardial effusion or wall-motion abnormalities. A random sample of exams wasselected for evaluation of variables accounting for disagreement. A sample of 768 patients was analyzed, 651(85%) in the referred group. Quality issues were reported in 5.8%, and the random Kappa for major HD between screening and standard echo was 0.51. The most frequent reasons for disagreement were: overestimation of mitral regurgitation (MR) (17.9%, N=138), left ventricular (LV) dysfunction (15.7%, N=121), aortic regurgitation (AR) (15.2%, N=117), LV hypertrophy (13.5%, N=104) and tricuspid regurgitation (12.7%, N=98). Misdiagnosis of mitral and aortic morphological abnormalities was observed in 12.4% and 3.0%, and underestimation of AR and MR occurred in 4.6% and 11.1%. Among 257 patients with suspected mild/moderate MR, 129 were reclassified to normal. In conclusion, although screening echo with task-shifting in PC is a promising tool in low-income areas, estimation of valve regurgitation and LV function and size account for considerable disagreement with standard exams.

Impact of a large-scale telemedicine network on emergency visits and hospital admissions during the coronavirus disease 2019 pandemic in Brazil: Data from the UNIMED-BH system.

INTRODUCTION: Triage by on-demand telemedicine is a strategy for healthcare surge control in the COVID-19 pandemic. We aimed to assess the impact of a large-scale COVID-19 telemedicine system on emergency department (ED) visits and all-cause and cardiovascular hospital admissions in Brazil. METHODS: From March 18, 2020-May 18, 2020 we evaluated the database of a cooperative private health insurance, with 1.28 million clients. The COVID-19 telemedicine system consisted of: a) mobile app, which redirects to teleconsultations if indicated; b) telemonitoring system, with regular phone calls to suspected/confirmed COVID-19 cases to monitor progression; c) emergency ambulance system (EAS), with internet phone triage and counselling. ED visits and hospital admissions were recorded, with diagnoses assessed by the Diagnosis Related Groups method. COVID-19 diagnosis and deaths were identified from the patients' registries, and outcomes assessed until June 1st. RESULTS: In 60 days, 24,354 patients accessed one of the telemedicine systems. The most frequently utilized was telemonitoring (16,717, 69%), followed by teleconsultation (13,357, 55%) and EAS (687, 3%). The rates of ED and hospital admissions were: telemonitoring 19.7% (3,296) and 4.7% (782); teleconsultation 17.3% (2,313) and 2.4% (318) and EAS: 55.9% (384) and 56.5% (388) patients. At total 4.1% (1,010) had hospital admissions, 36% (363) with respiratory diseases (44 requiring mechanical ventilation) and 4.4% (44) with cardiovascular diagnoses. Overall, 277 (1.1%) patients had confirmed COVID-19 diagnosis, and 160 (0.7%) died, 9 with COVID-19. CONCLUSION: Telemedicine resulted in low rates of ED visits and hospital admissions, suggesting positive impacts on healthcare utilization. Cardiovascular admissions were remarkably rare.