Prediction of atrial fibrillation in patients with cardiac dysfunctions†

Applications of Machine Learning in Ambulatory ECG

The ambulatory ECG (AECG) is an important diagnostic tool for many heart electrophysiology-related cases. AECG covers a wide spectrum of devices and applications. At the core of these devices and applications are the algorithms responsible for signal conditioning, ECG beat detection and classification, and event detections. Over the years, there has been huge progress for algorithm development and implementation thanks to great efforts by researchers, engineers, and physicians, alongside the rapid development of electronics and signal processing, especially machine learning (ML). The current efforts and progress in machine learning fields are unprecedented, and many of these ML algorithms have also been successfully applied to AECG applications. This review covers some key AECG applications of ML algorithms. However, instead of doing a general review of ML algorithms, we are focusing on the central tasks of AECG and discussing what ML can bring to solve the key challenges AECG is facing. The center tasks of AECG signal processing listed in the review include signal preprocessing, beat detection and classification, event detection, and event prediction. Each AECG device/system might have different portions and forms of those signal components depending on its application and the target, but these are the topics most relevant and of greatest concern to the people working in this area.

Electrocardiogram (ECG) for the Prediction of Incident Atrial Fibrillation: An Overview

Electrocardiograms (ECGs) have been employed to medically evaluate participants in population-based studies, and ECG-derived predictors have been reported for incident atrial fibrillation (AF). Here, we reviewed the status of ECG in predicting new-onset AF. We surveyed population-based studies and revealed ECG variables to be risk factors for incident AF. When available, the predictive values of each ECG risk marker were calculated. Both the atrium-related and ventricle-related ECG variables were risk factors for incident AF, with significant hazard risks (HRs) even after multivariate adjustments. The risk factors included P-wave indices (maximum P-wave duration, its dispersion or variation and P-wave morphology) and premature atrial contractions (PACs) or runs. In addition, left ventricular hypertrophy (LVH), ST-T abnormalities, intraventricular conduction delay, QTc interval and premature ventricular contractions (PVCs) or runs were a risk of incident AF. An HR of greater than 2.0 was observed in the upper 5th percentile of the P-wave durations, P-wave durations greater than 130 ms, P-wave morpholyg, PACs (PVCs) or runs, LVH, QTc and left anterior fascicular blocks. The sensitivity , specificity and the positive and negative predictive values were 3.6-53.8%, 61.7-97.9%, 2.9-61.7% and 77.4-97.7%, respectively. ECG variables are risk factors for incident AF. The correlation between the ECG-derived AF predictors, especially P-wave indices, and underlying diseases and the effects of the reversal of the ECG-derived predictors on incident AF by treatment of comorbidities require further study.

Number of P-wave fragmentations on P-SAECG correlates with infiltrated atrial fat

BACKGROUND

Although atrial fibrillation (AF) triggers are known, the underlying AF substrate is less well understood. The goal of our study was to explore correlations between electrophysiological and structural characteristics of atria in patients with paroxysmal AF and individuals at AF risk.

METHODS

Patients in sinus rhythm (N = 90; age 57 ± 10 year; 55 men [63.2%]) with structural heart disease and paroxysmal AF (n = 12 [13%]), or with AF risk factors and LVEF > 35% (n = 78), underwent SAECG and cardiac magnetic resonance study. Interatrial and epicardial fat was analyzed with a Dark-blood DIR-prepared Fat-Water-separated sequence in the horizontal longitudinal axis. All local P-wave extrema were identified on SAECG leads during sinus rhythm. A P-wave fragmentation (Pf) was defined as an absolute difference between adjacent extrema which was above three standard deviations of noise, and was normalized by the duration of the P wave in the corresponding lead.

RESULTS

The Pf was greater on the filtered than on the unfiltered P-SAECG signal (13.1 ± 3.8 vs. 3.4 ± 1.2; P < 0.0001). Pf was the greatest on the Y lead (13.0 ± 3.5 on Y lead vs. 12.1 ± 3.4 on Z lead; P = 0.003. Pf on Z lead correlated with interatrial fat index (r = 0.544; P = 0.001). Epicardial fat significantly correlated with body mass index (BMI; r = 0.302; P = 0.015). After adjustment for BMI, left atrium (LA) size, epicardial fat, and interatrial septum width, interatrial fat independently associated with the Pf on Z lead (β-coefficient 0.009 [95%CI 0.0003-0.019]; P = 0.043).

CONCLUSIONS

Infiltrated atrial fat correlates with discontinuous conduction on posterior LA wall and represents AF early substrate.

Incidence and predictors of asymptomatic atrial fibrillation in patients older than 70 years with complete atrioventricular block and dual chamber pacemaker implantation

AIM

To evaluate predictors of asymptomatic atrial fibrillation in patients older than 70 years with complete atrioventricular (AV) block, normal left ventricular systolic function, and implanted dual chamber (DDD) pacemaker.

METHODS

Hundred and eighty six patients with complete AV block were admitted over one year to the Sisters of Mercy University Hospital. The study recruited patients older than 70 years, with no history of atrial fibrillation, heart failure, or reduced left ventricular systolic function. All the patients were implanted with the same pacemaker. Out of 103 patients who were eligible for the study, 81 (78%) were evaluated. Among those 81 (78%) were evaluated. Eighty one (78%) patients were evaluated. Follow-up time ranged from 12 to 33 months (average ±standard deviation 23 ± 5 months). Primary end-point was asymptomatic atrial fibrillation occurrence recorded by the pacemaker. Atrial fibrillation occurrence was defined as atrial high rate episodes (AHRE) lasting >5 minutes. Binary logistic regression was used to identify the predictors of development of asymptomatic atrial fibrillation. Results. The 81 patients were stratified into two groups depending on the presence of AHRE lasting >5 minutes (group 1 had AHRE>5 minutes and group 2 AHRE<5 minutes). AHRE lasting >5 minutes were detected in 49 (60%) patients after 3 months and in 53 (65%) patients after 18 moths. After 3 months, only hypertension (odds ratio [OR], 17.63; P = 0.020) was identified as a predictor of asymptomatic atrial fibrillation. After 18 months, hypertension (OR, 14.0; P = 0.036), P wave duration >100 ms in 12 lead ECG (OR, 16.5; P = 0.001), and intracardial atrial electrogram signal amplitude >4 mV (OR, 4.27; P = 0.045) were identified as predictors of atrial fibrillation.

CONCLUSION

In our study population, hypertension was the most robust and constant predictor of asymptomatic atrial fibrillation after 3 months, while P wave duration >100 ms in 12-lead ECG and intracardial atrial signal amplitude were predictors after 18 months.

Improvement of P wave dispersion after cardiac resynchronization therapy for heart failure

OBJECTIVE

The purpose of this study is to investigate the effect of cardiac resynchronization therapy (CRT) on P wave maximum duration (PWM) and P wave dispersion (PWD) in patients with advanced heart failure.

METHODS

Forty-six patients (33 men; mean age, 60 +/- 11 years) with CRT were enrolled in the present study. PWM and PWD were measured using 12-lead surface electrocardiography (ECG) at a paper speed of 50 mm/s and 20 mm/mV. Serial ECG, echocardiography, clinical assessment, and device interrogations were performed at baseline and 3 months after CRT.

RESULTS

After 3 months of follow-up, PWM and PWD values were significantly decreased (129.6 +/- 11.3 to 120.7 +/- 10.7 milliseconds, P < .001; 42.6 +/- 8.0 to 32.3 +/- 10.1 milliseconds; P < .001, respectively). It showed a significant reduction in left atrial diameter (LAD) (46.5 +/- 5.2 to 44.9 +/- 5.6 mm, P = .021) and an improvement in left ventricular ejection fraction (LVEF) (29.0% +/- 7.5% to 36.2% +/- 8.0%, P < .001). The decrease of PWM and PWD was positively correlated with the reduction of LAD and negatively correlated with the improvement of LVEF. The reduction in atrial fibrillation burden was observed after 3 months of follow-up.

CONCLUSIONS

Cardiac resynchronization therapy decreases PWM and PWD along with an improvement of LVEF and a reduction of LAD. Further studies are needed to evaluate the clinical implications of decrease of PWD on prevention of atrial fibrillation.