Thyroid dysfunction and exudative age-related macular degeneration - A longitudinal nationwide registry-based cohort study.

PURPOSE: The association between thyroid dysfunction and exudative age-related macular degeneration (AMD) is unknown. METHODS: In this Danish longitudinal nationwide registry-based cohort study we included all Danish residents aged 50-100 between 2008 and 2018. Using the Danish national registries, we studied the association between thyroid dysfunction and exudative AMD. Thyroid dysfunction was classified as two consecutive redeemed prescriptions of thyroid hormones (hypothyroidism) or anti-thyroid medication (hyperthyroidism). Exudative AMD was classified as an ICD diagnosis of AMD and a code for anti-VEGF treatment. All patients are treated for exudative AMD in a hospital in Denmark, and we therefore have complete registration of this patient group. RESULTS: We included 2 087 305 individuals, of which 1 072 567 (51.4%) were women; 59 318 (2.8%) had hypothyroidism, and 33 922 (1.6%) had hyperthyroidism. During a median follow-up of 11 years, 26 998 (1.3%) people developed exudative AMD. Hypothyroidism (adjusted hazard ratio [HR]: 1.17; 95% confidence interval [CI] 1.10-1.25; p < 0.001) and hyperthyroidism (HR: 1.23; 95% CI:1.13-1.34; p < 0.001) were both associated with the development of exudative AMD. The age-stratified analyses yielded similar results to the main analyses, except that the risks were exaggerated in the older part of the population. CONCLUSION: This is the first longitudinal nationwide study showing that both hypo- and hyperthyroidism are associated with an increased risk of exudative AMD. AMD is a quantitative problem in the population and our findings could have a public health impact. Further studies are needed to study the underlying mechanisms of the association.

Electrocardiographic markers in patients with type 2 diabetes and the role of diabetes duration.

BACKGROUND: The association between type 2 diabetes and electrocardiographic (ECG) markers are incompletely explored and the dependence on diabetes duration is largely unknown. We aimed to investigate the electrocardiographic (ECG) changes associated with type 2 diabetes over time. METHODS: In this cross-sectional study, we matched people with type 2 diabetes 1:1 on sex, age, and body mass index with people without diabetes from the general population. We regressed ECG markers with the presence of diabetes and the duration of clinical diabetes, respectively, adjusted for sex, age, body mass index, smoking, heart rate, diabetes medication, renal function, hypertension, and myocardial infarction. RESULTS: We matched 988 people with type 2 diabetes (332, 34% females) with as many controls. Heart rate was 8 bpm higher (p < 0.001) in people with vs. without type 2 diabetes, but the difference declined with increasing diabetes duration. For most depolarization markers, the difference between people with and without type 2 diabetes increased progressively with diabetes duration. On average, R-wave amplitude was 6 mm lower in lead V5 (p < 0.001), P-wave duration was 5 ms shorter (p < 0.001) and QRS duration was 3 ms (p = 0.03). Among repolarization markers, T-wave amplitude (measured in V5) was lower in patients with type 2 diabetes (1 mm lower, p < 0.001) and the QRS-T angle was 10 degrees wider (p = 0.002). We observed no association between diabetes duration and repolarization markers. CONCLUSIONS: Type 2 diabetes was independently associated with electrocardiographic depolarization and repolarization changes. Differences in depolarization markers, but not repolarization markers, increased with increasing diabetes duration.

Ventricular tachycardia and in-hospital mortality in the intensive care unit.

Background: Continuous electrocardiographic (ECG) monitoring is used to identify ventricular tachycardia (VT), but false alarms occur frequently. Objective: The purpose of this study was to assess the rate of 30-day in-hospital mortality associated with VT alerts generated from bedside ECG monitors to those from a new algorithm among intensive care unit (ICU) patients. Methods: We conducted a retrospective cohort study in consecutive adult ICU patients at an urban academic medical center and compared current bedside monitor VT alerts, VT alerts from a new-unannotated algorithm, and true-annotated VT. We used survival analysis to explore the association between VT alerts and mortality. Results: We included 5679 ICU admissions (mean age 58 ± 17 years; 48% women), 503 (8.9%) experienced 30-day in-hospital mortality. A total of 30.1% had at least 1 current bedside monitor VT alert, 14.3% had a new-unannotated algorithm VT alert, and 11.6% had true-annotated VT. Bedside monitor VT alert was not associated with increased rate of 30-day mortality (adjusted hazard ratio [aHR] 1.06; 95% confidence interval [CI] 0.88-1.27), but there was an association for VT alerts from our new-unannotated algorithm (aHR 1.38; 95% CI 1.12-1.69) and true-annotated VT(aHR 1.39; 95% CI 1.12-1.73). Conclusion: Unannotated and annotated-true VT were associated with increased rate of 30-day in-hospital mortality, whereas current bedside monitor VT was not. Our new algorithm may accurately identify high-risk VT; however, prospective validation is needed.

3D-electroanatomical mapping of the left atrium and catheter-based pulmonary vein isolation in pigs: A practical guide.

Aim: To propose a standardized workflow for 3D-electroanatomical mapping guided pulmonary vein isolation in pigs. Materials and methods: Danish female landrace pigs were anaesthetized. Ultrasound-guided puncture of both femoral veins was performed and arterial access for blood pressure measurement established. Fluoroscopy- and intracardiac ultrasound-guided passage of the patent foramen ovale or transseptal puncture was performed. Then, 3D-electroanatomical mapping of the left atrium was conducted using a high-density mapping catheter. After mapping all pulmonary veins, an irrigated radiofrequency ablation catheter was used to perform ostial ablation to achieve electrical pulmonary vein isolation. Entrance- and exit-block were confirmed and re-assessed after a 20-min waiting period. Lastly, animals were sacrificed to perform left atrial anatomical gross examination. Results: We present data from 11 consecutive pigs undergoing pulmonary vein isolation. Passage of the fossa ovalis or transseptal puncture was uneventful and successful in all animals. Within the inferior pulmonary trunk 2-4 individual veins as well as 1-2 additional left and right pulmonary veins could be cannulated. Electrical isolation by point-by-point ablation of all targeted veins was successful. However, pitfalls including phrenic nerve capture during ablation, ventricular arrhythmias during antral isolation close to the mitral valve annulus and difficulties in accessing right pulmonary veins were encountered. Conclusion: Fluoroscopy- and intracardiac ultrasound-guided transseptal puncture, high-density electroanatomical mapping of all pulmonary veins and complete electrical pulmonary vein isolation can be achieved reproducibly and safely in pigs when using current technologies and a step-by-step approach.

Genetic architecture of spatial electrical biomarkers for cardiac arrhythmia and relationship with cardiovascular disease.

The 3-dimensional spatial and 2-dimensional frontal QRS-T angles are measures derived from the vectorcardiogram. They are independent risk predictors for arrhythmia, but the underlying biology is unknown. Using multi-ancestry genome-wide association studies we identify 61 (58 previously unreported) loci for the spatial QRS-T angle (N = 118,780) and 11 for the frontal QRS-T angle (N = 159,715). Seven out of the 61 spatial QRS-T angle loci have not been reported for other electrocardiographic measures. Enrichments are observed in pathways related to cardiac and vascular development, muscle contraction, and hypertrophy. Pairwise genome-wide association studies with classical ECG traits identify shared genetic influences with PR interval and QRS duration. Phenome-wide scanning indicate associations with atrial fibrillation, atrioventricular block and arterial embolism and genetically determined QRS-T angle measures are associated with fascicular and bundle branch block (and also atrioventricular block for the frontal QRS-T angle). We identify potential biology involved in the QRS-T angle and their genetic relationships with cardiovascular traits and diseases, may inform future research and risk prediction.

Accuracy of continuous photoplethysmography-based 1 min mean heart rate assessment during atrial fibrillation.

AIMS: Although mobile health tools using photoplethysmography (PPG) technology have been validated for the detection of atrial fibrillation (AF), their utility for heart rate assessment during AF remains unclear. Therefore, we aimed to evaluate the accuracy of continuous PPG-based 1 min mean heart rate assessment during AF. METHODS AND RESULTS: Persistent AF patients were provided with Holter electrocardiography (ECG) (for ≥24 h) simultaneously with a PPG-equipped smartwatch. Both the PPG-based smartwatch and Holter ECG automatically and continuously monitored patients' heart rate/rhythm. ECG and PPG recordings were synchronized and divided into 1 min segments, from which a PPG-based and an ECG-based average heart rate estimation were extracted. In total, 47 661 simultaneous ECG and PPG 1 min heart rate segments were analysed in 50 patients (34% women, age 73 ± 8 years). The agreement between ECG-determined and PPG-determined 1 min mean heart rate was high [root mean squared error (RMSE): 4.7 bpm]. The 1 min mean heart rate estimated using PPG was accurate within ±10% in 93.7% of the corresponding ECG-derived 1 min mean heart rate segments. PPG-based 1 min mean heart rate estimation was more often accurate during night-time (97%) than day-time (91%, P < 0.001) and during low levels (96%) compared to high levels of motion (92%, P < 0.001). A neural network with a 10 min history of the recording did not further improve the PPG-based 1 min mean heart rate assessment [RMSE: 4.4 (95% confidence interval: 3.5-5.2 bpm)]. Only chronic heart failure was associated with a lower agreement between ECG-derived and PPG-derived 1 min mean heart rates (P = 0.040). CONCLUSION: During persistent AF, continuous PPG-based 1 min mean heart rate assessment is feasible in 60% of the analysed period and shows high accuracy compared with Holter ECG for heart rates <110 bpm.

Association between primary care electrocardiogram markers and Alzheimer's disease.

OBJECTIVE: The association between common electrocardiogram (ECG) markers and Alzheimer's disease has been scarcely investigated, and it is unknown if ECG markers can improve risk prediction. Thus, we aimed to examine the association between common ECG markers and Alzheimer's disease in a large population. METHODS: We studied the association between ECG markers and Alzheimer's disease using Cox models with adjustment for age, sex, and comorbidities using a large primary care population of patients aged 60 years or more. RESULTS: We followed 172,236 subjects for a median of 7.5 years. Increased PR interval (hazard ratio for PR > 188 ms: 0.76 [95% confidence interval: 0.69-0.83, p < 0.001) and increased QTc interval (hazard ratio for QTc = [426;439]: 0.90 [0.83-0.98], p = 0.02) were associated with a decreased rate of Alzheimer's disease. A positive Sokolow-Lyon index >35 mm (1.22 [1.13-1.33], p < 0.001) and increased T-wave amplitude >4.1 mm (1.15 [1.04-1.27]) were associated with an increased rate of Alzheimer's disease. Upon addition of ECG markers to a reference model, 10-year prediction area under the receiver-operator characteristics curve (AUC) improved by 0.39 [0.06-0.67] %-points. The 10-year absolute risk of Alzheimer's disease was 6.5% and 5.2% for an 82-year old female and a male, respectively, with a favorable ECG, and 12% and 9.2%, respectively, with an unfavorable ECG, almost twice as high. CONCLUSIONS: We identified several common ECG markers which were associated with Alzheimer's disease, and which improved risk prediction for Alzheimer's disease.

Intramural Purkinje fibers facilitate rapid ventricular activation in the equine heart.

BACKGROUND: The Purkinje fibers convey the electrical impulses at much higher speed than the working myocardial cells. Thus, the distribution of the Purkinje network is of paramount importance for the timing and coordination of ventricular activation. The Purkinje fibers are found in the subendocardium of all species of mammals, but some mammals also possess an intramural Purkinje fiber network that provides for relatively instantaneous, burst-like activation of the entire ventricular wall, and gives rise to an rS configuration in lead II of the ECG. AIM: To relate the topography of the horse heart and the distribution and histology of the conduction system to the pattern of ventricular activation as a mechanism for the unique electrical axis of the equine heart. METHODS: The morphology and distribution of the cardiac conduction system was determined by histochemistry. The electrical activity was measured using ECG in the Einthoven and orthogonal configuration. RESULTS: The long axis of the equine heart is close to vertical. Outside the nodal regions the conduction system consisted of Purkinje fibers connected by connexin 43 and long, slender parallel running transitional cells. The Purkinje fiber network extended deep into the ventricular walls. ECGs recorded in an orthogonal configuration revealed a mean electrical axis pointing in a cranial-to-left direction indicating ventricular activation in an apex-to-base direction. CONCLUSION: The direction of the mean electrical axis in the equine heart is determined by the architecture of the intramural Purkinje network, rather than being a reflection of ventricular mass.

A randomized, double-blind, crossover study of the effect of the fluoroquinolone moxifloxacin on glucose levels and insulin sensitivity in young men and women.

AIM: The voltage-gated potassium channel Kv 11.1 is important for repolarizing the membrane potential in excitable cells such as myocytes, pancreatic α- and ß-cells. Moxifloxacin blocks the Kv 11.1 channel and increases the risk of hypoglycaemia in patients with diabetes. We investigated glucose regulation and secretion of glucoregulatory hormones in young people with and without moxifloxacin, a drug known to block the Kv 11.1 channel. MATERIALS AND METHODS: The effect of moxifloxacin (800 mg/day for 4 days) or placebo on glucose regulation was assessed in a randomized, double-blind, crossover study of young men and women (age 20-40 years and body mass index 18.5-27.5 kg/m2 ) without chronic disease, using 6-h oral glucose tolerance tests and continuous glucose monitoring. RESULTS: Thirty-eight participants completed the study. Moxifloxacin prolonged the QTcF interval and increased heart rate. Hypoglycaemia was more frequently observed with moxifloxacin, both during the 8 days of continuous glucose monitoring and during the oral glucose tolerance tests. Hypoglycaemia questionnaire scores were higher after intake of moxifloxacin. Moxifloxacin reduced the early plasma-glucose response (AUC0-30 min ) by 7% (95% CI: -9% to -4%, p < .01), and overall insulin response (AUC0-360 min ) decreased by 18% (95% CI: -24% to -11%, p < .01) and plasma glucagon increased by 17% (95% CI: 4%-33%, p = .03). Insulin sensitivity calculated as the Matsuda index increased by 11%, and MISI, an index of muscle insulin sensitivity, increased by 34%. CONCLUSIONS: In young men and women, moxifloxacin, a drug known to block the Kv 11.1 channel, increased QT interval, decreased glucose levels and was associated with increased muscle insulin sensitivity and more frequent episodes of hypoglycaemia.

Genetic analyses of the electrocardiographic QT interval and its components identify additional loci and pathways.

The QT interval is an electrocardiographic measure representing the sum of ventricular depolarization and repolarization, estimated by QRS duration and JT interval, respectively. QT interval abnormalities are associated with potentially fatal ventricular arrhythmia. Using genome-wide multi-ancestry analyses (>250,000 individuals) we identify 177, 156 and 121 independent loci for QT, JT and QRS, respectively, including a male-specific X-chromosome locus. Using gene-based rare-variant methods, we identify associations with Mendelian disease genes. Enrichments are observed in established pathways for QT and JT, and previously unreported genes indicated in insulin-receptor signalling and cardiac energy metabolism. In contrast for QRS, connective tissue components and processes for cell growth and extracellular matrix interactions are significantly enriched. We demonstrate polygenic risk score associations with atrial fibrillation, conduction disease and sudden cardiac death. Prioritization of druggable genes highlight potential therapeutic targets for arrhythmia. Together, these results substantially advance our understanding of the genetic architecture of ventricular depolarization and repolarization.

Mobile app-based symptom-rhythm correlation assessment in patients with persistent atrial fibrillation.

BACKGROUND: The assessment of symptom-rhythm correlation (SRC) in patients with persistent atrial fibrillation (AF) is challenging. Therefore, we performed a novel mobile app-based approach to assess SRC in persistent AF. METHODS: Consecutive persistent AF patients planned for electrical cardioversion (ECV) used a mobile app to record a 60-s photoplethysmogram (PPG) and report symptoms once daily and in case of symptoms for four weeks prior and three weeks after ECV. Within each patient, SRC was quantified by the SRC-index defined as the sum of symptomatic AF recordings and asymptomatic non-AF recordings divided by the sum of all recordings. RESULTS: Of 88 patients (33% women, age 68 ± 9 years) included, 78% reported any symptoms during recordings. The overall SRC-index was 0.61 (0.44-0.79). The study population was divided into SRC-index tertiles: low (<0.47), medium (0.47-0.73) and high (≥0.73). Patients within the low (vs high) SRC-index tertile had more often heart failure and diabetes mellitus (both 24.1% vs 6.9%). Extrasystoles occurred in 19% of all symptomatic non-AF PPG recordings. Within each patient, PPG recordings with the highest (vs lowest) tertile of pulse rates conferred an increased risk for symptomatic AF recordings (odds ratio [OR] 1.26, 95% coincidence interval [CI] 1.04-1.52) and symptomatic non-AF recordings (OR 2.93, 95% CI 2.16-3.97). Pulse variability was not associated with reported symptoms. CONCLUSIONS: In patients with persistent AF, SRC is relatively low. Pulse rate is the main determinant of reported symptoms. Further studies are required to verify whether integrating mobile app-based SRC assessment in current workflows can improve AF management.

Associations between primary care electrocardiography and non-Alzheimer dementia.

OBJECTIVES: To determine whether electrocardiogram (ECG) markers are associated with incident non-Alzheimer's dementia (non-AD) and whether these markers also improve risk prediction for non-AD. MATERIALS AND METHODS: We retrospectively included 170,605 primary care patients aged 60 years or older referred for an ECG by their general practitioner and followed them for a median of 7.6 years. Using Cox regression, we reported hazard ratios (HRs) for electrocardiogram markers. Subsequently, we evaluated if addition of these electrocardiogram markers to a clinical model improved risk prediction for non-AD using change in area under the receiver-operator characteristics curve (AUC). RESULTS: The 5-year cumulative incidence of non-AD was 3.4 %. Increased heart rate (HR=1.06 pr. 10 bpm [95% confidence interval: 1.04-1.08], p<0.001), shorter QRS duration (HR=1.07 pr. 10 ms [1.05-1.09], p<0.001), elevated J-amplitude (HR=1.16 pr. mm [1.08-1.24], p<0.001), decreased T-peak amplitude (HR=1.02 pr. mm [1.01-1.04], p=0.002), and increased QTc (HR=1.08 pr. 20 ms [1.05-1.10], p<0.001) were associated with an increased rate of non-AD. Atrial fibrillation on the ECG (HR=1.18 [1.08-1.28], p<0.001) Sokolow-Lyon index > 35 mm (HR=1.31 [1.18-1.46], p<0.001) and borderline (HR=1.18 [1.11-1.26], p<0.001) or abnormal (HR=1.40 [1.27-1.55], p<0.001) QRS-T angle were also associated with an increased rate of non-AD. Upon addition of ECG markers to the Cox model, 5-year and 10-year C-statistic (AUC) improved significantly (delta-AUC, 0.36 [0.18-0.50] and 0.20 [0.03-0.35] %-points, respectively). CONCLUSIONS: ECG markers typical of an elevated cardiovascular risk profile were associated with non-AD and improved both 5-year and 10-year risk predictions for non-AD.

Electrocardiographic characteristics of trained and untrained standardbred racehorses.

BACKGROUND: Long-term exercise induces cardiac remodeling that potentially influences the electrical properties of the heart. HYPOTHESIS/OBJECTIVES: We assessed whether training alters cardiac conduction in Standardbred racehorses. ANIMALS: Two hundred one trained and 52 untrained Standardbred horses. METHODS: Cross-sectional study. Resting ECG recordings were analyzed to assess heart rate (HR) along with standard ECG parameters and for identification of atrial and ventricular arrhythmias. An electrophysiological study was performed in 13 horses assessing the effect of training on sinoatrial (SA) and atrioventricular (AV) nodal function by sinus node recovery time (SNRT) and His signal recordings. Age and sex adjustments were implemented in multiple and logistic regression models for comparison. RESULTS: Resting HR in beats per minute (bpm) was lower in trained vs untrained horses (mean, 30.8 ± 2.6 bpm vs 32.9 ± 4.2 bpm; P = .001). Trained horses more often displayed second-degree atrioventricular block (2AVB; odds ratio, 2.59; P = .04). No difference in SNRT was found between groups (n = 13). Mean P-A, A-H, and H-V intervals were 71 ± 20, 209 ± 41, and 134 ± 41 ms, respectively (n = 7). We did not detect a training effect on AV-nodal conduction intervals. His signals were present in 1 horse during 2AVB with varying H-V interval preceding a blocked beat. CONCLUSIONS AND CLINICAL IMPORTANCE: We identified decreased HR and increased frequency of 2AVB in trained horses. In 5 of 7 horses, His signal recordings had variable H-V intervals within each individual horse, providing novel insight into AV conduction in horses.

Artificial intelligence for the detection, prediction, and management of atrial fibrillation.

The present article reviews the state of the art of machine learning algorithms for the detection, prediction, and management of atrial fibrillation (AF), as well as of the development and evaluation of artificial intelligence (AI) in cardiology and beyond. Today, AI detects AF with a high accuracy using 12-lead or single-lead electrocardiograms or photoplethysmography. The prediction of paroxysmal or future AF currently operates at a level of precision that is too low for clinical use. Further studies are needed to determine whether patient selection for interventions may be possible with machine learning.

Electrocardiography in euthyroid individuals: a Danish general population study.

BACKGROUND: Thyroid hormones within the euthyroid range have been linked to mortality and differences in heart rate. However, some relations between thyroid hormone concentration and various electrocardiographic measurements remain unassessed. We aimed to investigate the association between thyroid hormone concentrations within the euthyroid range and different electrocardiographic markers in people free of thyroid disease. METHODS: We obtained electrocardiograms (ECG) and blood samples of free T4, total T3, and thyrotropin (TSH) in 20,852 subjects from the general population (the GESUS study). Relations between concentrations of TSH, free T4, and total T3 and heart rate, QTc, QRS duration, PR interval, P-wave duration and T-wave morphology were assessed in a multivariate adjusted linear model stratified by sex. RESULTS: Roughly half of the 18,046 included participants with thyroid hormone measurements within euthyroid range were men, and the average age was 56 years. Heart rate increased with concentrations of T3 (6.4 bpm/nM, P<0.001 in women and 5.3 bpm/nM, P<0.001 in men) and T4 (3.7 bpm/10pM, P<0.001 in women and 3.1 bpm/10pM, P<0.001 in men). We found no relation between TSH and heart rate. PR interval and QRS duration decreased with higher concentrations of T3 (all P<0.01). QTc increased with higher concentrations of T4 in men (5 ms/10pM), and T waves were flatter, more asymmetric, and more often had notches with higher concentrations of T4 (all P≤0.01). CONCLUSIONS: Thyroid hormone concentrations within the euthyroid range in people free of thyroid disease were associated with changes in the electrocardiogram in a general population.

Obesity Partially Mediates the Diabetogenic Effect of Lowering LDL Cholesterol.

OBJECTIVE: LDL cholesterol (LDLc)-lowering drugs modestly increase body weight and type 2 diabetes risk, but the extent to which the diabetogenic effect of lowering LDLc is mediated through increased BMI is unknown. RESEARCH DESIGN AND METHODS: We conducted summary-level univariable and multivariable Mendelian randomization (MR) analyses in 921,908 participants to investigate the effect of lowering LDLc on type 2 diabetes risk and the proportion of this effect mediated through BMI. We used data from 92,532 participants from 14 observational studies to replicate findings in individual-level MR analyses. RESULTS: A 1-SD decrease in genetically predicted LDLc was associated with increased type 2 diabetes odds (odds ratio [OR] 1.12 [95% CI 1.01, 1.24]) and BMI (ß = 0.07 SD units [95% CI 0.02, 0.12]) in univariable MR analyses. The multivariable MR analysis showed evidence of an indirect effect of lowering LDLc on type 2 diabetes through BMI (OR 1.04 [95% CI 1.01, 1.08]) with a proportion mediated of 38% of the total effect (P = 0.03). Total and indirect effect estimates were similar across a number of sensitivity analyses. Individual-level MR analyses confirmed the indirect effect of lowering LDLc on type 2 diabetes through BMI with an estimated proportion mediated of 8% (P = 0.04). CONCLUSIONS: These findings suggest that the diabetogenic effect attributed to lowering LDLc is partially mediated through increased BMI. Our results could help advance understanding of adipose tissue and lipids in type 2 diabetes pathophysiology and inform strategies to reduce diabetes risk among individuals taking LDLc-lowering medications.

DeepFake electrocardiograms using generative adversarial networks are the beginning of the end for privacy issues in medicine.

Recent global developments underscore the prominent role big data have in modern medical science. But privacy issues constitute a prevalent problem for collecting and sharing data between researchers. However, synthetic data generated to represent real data carrying similar information and distribution may alleviate the privacy issue. In this study, we present generative adversarial networks (GANs) capable of generating realistic synthetic DeepFake 10-s 12-lead electrocardiograms (ECGs). We have developed and compared two methods, named WaveGAN* and Pulse2Pulse. We trained the GANs with 7,233 real normal ECGs to produce 121,977 DeepFake normal ECGs. By verifying the ECGs using a commercial ECG interpretation program (MUSE 12SL, GE Healthcare), we demonstrate that the Pulse2Pulse GAN was superior to the WaveGAN* to produce realistic ECGs. ECG intervals and amplitudes were similar between the DeepFake and real ECGs. Although these synthetic ECGs mimic the dataset used for creation, the ECGs are not linked to any individuals and may thus be used freely. The synthetic dataset will be available as open access for researchers at OSF.io and the DeepFake generator available at the Python Package Index (PyPI) for generating synthetic ECGs. In conclusion, we were able to generate realistic synthetic ECGs using generative adversarial neural networks on normal ECGs from two population studies, thereby addressing the relevant privacy issues in medical datasets.

Atrial fibrillation burden and cognitive decline in elderly patients undergoing continuous monitoring.

AIMS: To study the relationship between subclinical atrial fibrillation (AF) and changes in cognitive function in a large cohort of individuals with stroke risk factors. METHODS: Individuals with no prior AF diagnosis but with risk factors for stroke were recruited to undergo annual cognitive assessment with the Montreal Cognitive Assessment (MoCA) along with implantable loop recorder (ILR) monitoring for AF for 3 years. If AF episodes lasting ≥6 minutes were detected, oral anticoagulation (OAC) treatment was initiated. RESULTS: A total of 1194 participants (55.2 % men, mean age 74.5 (±3.9)) had a combined duration of heart rhythm monitoring of ≈1.3 million days. Among these, 339 participants (28.3%) had adjudicated AF, with a median AF burden of 0.072% (0.02, 0.39), and 324 (96%) initiated OAC. When stratifying the participants into AF burden groups (No AF, AFlow (AF burden <0.25%), and AFhigh, (AF burden >0.25%)), only participants in the AFlow group had a decrease in MoCA score over time (P = .03), although this was not significant after adjustment for stroke risk factors. A subgroup analysis of 175 participants (14.6%) with a MoCA <26 at 3 years found no association to AF diagnosis or burden. CONCLUSIONS: In a high-risk population, subclinical AF detected by continuous monitoring and subsequently treated with OAC was not associated with a significant change in MoCA score over a 3-year period.

Explaining deep neural networks for knowledge discovery in electrocardiogram analysis.

Deep learning-based tools may annotate and interpret medical data more quickly, consistently, and accurately than medical doctors. However, as medical doctors are ultimately responsible for clinical decision-making, any deep learning-based prediction should be accompanied by an explanation that a human can understand. We present an approach called electrocardiogram gradient class activation map (ECGradCAM), which is used to generate attention maps and explain the reasoning behind deep learning-based decision-making in ECG analysis. Attention maps may be used in the clinic to aid diagnosis, discover new medical knowledge, and identify novel features and characteristics of medical tests. In this paper, we showcase how ECGradCAM attention maps can unmask how a novel deep learning model measures both amplitudes and intervals in 12-lead electrocardiograms, and we show an example of how attention maps may be used to develop novel ECG features.