Dataset Information

Machine learning enables noninvasive prediction of atrial fibrillation driver location and acute pulmonary vein ablation success using the 12-lead ECG.

ABSTRACT:

Background

Atrial fibrillation (AF) is the most common supraventricular arrhythmia, characterized by disorganized atrial electrical activity, maintained by localized arrhythmogenic atrial drivers. Pulmonary vein isolation (PVI) allows to exclude PV-related drivers. However, PVI is less effective in patients with additional extra-PV arrhythmogenic drivers.

Objectives

To discriminate whether AF drivers are located near the PVs vs extra-PV regions using the noninvasive 12-lead electrocardiogram (ECG) in a computational and clinical framework, and to computationally predict the acute success of PVI in these cohorts of data.

Methods

AF drivers were induced in 2 computerized atrial models and combined with 8 torso models, resulting in 1128 12-lead ECGs (80 ECGs with AF drivers located in the PVs and 1048 in extra-PV areas). A total of 103 features were extracted from the signals. Binary decision tree classifier was trained on the simulated data and evaluated using hold-out cross-validation. The PVs were subsequently isolated in the models to assess PVI success. Finally, the classifier was tested on a clinical dataset (46 patients: 23 PV-dependent AF and 23 with additional extra-PV sources).

Results

The classifier yielded 82.6% specificity and 73.9% sensitivity for detecting PV drivers on the clinical data. Consistency analysis on the 46 patients resulted in 93.5% results match. Applying PVI on the simulated AF cases terminated AF in 100% of the cases in the PV class.

Conclusion

Machine learning-based classification of 12-lead-ECG allows discrimination between patients with PV drivers vs those with extra-PV drivers of AF. The novel algorithm may aid to identify patients with high acute success rates to PVI.

SUBMITTER: Luongo G

PROVIDER: S-EPMC8053175 | biostudies-literature | 2021 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine learning enables noninvasive prediction of atrial fibrillation driver location and acute pulmonary vein ablation success using the 12-lead ECG.

Luongo Giorgio G Azzolin Luca L Schuler Steffen S Rivolta Massimo W MW Almeida Tiago P TP Martínez Juan P JP Soriano Diogo C DC Luik Armin A Müller-Edenborn Björn B Jadidi Amir A Dössel Olaf O Sassi Roberto R Laguna Pablo P Loewe Axel A

Cardiovascular digital health journal 20210401 2

<h4>Background</h4>Atrial fibrillation (AF) is the most common supraventricular arrhythmia, characterized by disorganized atrial electrical activity, maintained by localized arrhythmogenic atrial drivers. Pulmonary vein isolation (PVI) allows to exclude PV-related drivers. However, PVI is less effective in patients with additional extra-PV arrhythmogenic drivers.<h4>Objectives</h4>To discriminate whether AF drivers are located near the PVs vs extra-PV regions using the noninvasive 12-lead electr ...[more]

PMID: 33899043

Similar Datasets

Project description:ObjectiveTo determine the accuracy of QT measurement in a smartphone-operated, single-lead ECG (1L-ECG) device (AliveCor KardiaMobile 1L).DesignCross-sectional, within-patient diagnostic validation study.Setting/participantsPatients underwent a 12-lead ECG (12L-ECG) for any non-acute indication in primary care, April 2017-July 2018.InterventionSimultaneous recording of 1L-ECGs and 12L-ECGs with blinded manual QT assessment. OUTCOMES OF INTEREST: (1) Difference in QT interval in milliseconds (ms) between the devices; (2) measurement agreement between the devices (excellent agreement <20 ms and clinically acceptable agreement <40 ms absolute difference); (3) sensitivity and specificity for detection of extreme QTc (short (≤340 ms) or long (≥480 ms)), on 1L-ECGs versus 12L-ECGs as reference standard. In case of significant discrepancy between lead I/II of 12L-ECGs and 1L-ECGs, we developed a correction tool by adding the difference between QT measurements of 12L-ECG and 1L-ECGs.Results250 ECGs of 125 patients were included. The mean QTc interval, using Bazett's formula (QTcB), was 393±25 ms (mean±SD) in 1L-ECGs and 392±27 ms in lead I of 12L-ECGs, a mean difference of 1±21 ms, which was not statistically different (paired t-test (p=0.51) and Bland Altman method (p=0.23)). In terms of agreement between 1L-ECGs and lead I, QTcB had excellent agreement in 66.9% and clinically acceptable agreement in 93.4% of observations. The sensitivity and specificity of detecting extreme QTc were 0% and 99.2%, respectively. The comparison of 1L-ECG QTcB with lead II of 12L-ECGs showed a significant difference (p=<0.01), but when using a correction factor (+9 ms) this difference was cancelled (paired t-test (p=0.43) or Bland Altman test (p=0.57)). Moreover, it led to improved rates of excellent (71.3%) and clinically acceptable (94.3%) agreement.ConclusionSmartphone-operated 1L-ECGs can be used to accurately measure the QTc interval compared with simultaneously obtained 12L-ECGs in a primary care population. This may provide an opportunity for monitoring the effects of potential QTc-prolonging medications.

Project description:BackgroundObservational studies suggest that electrocardiogram (ECG) indices might be influenced by obesity and other anthropometric measures, though it is difficult to infer causal relationships based on observational data due to risk of residual confounding. We utilized mendelian randomization (MR) to explore causal relevance of multiple anthropometric measures on P-wave duration (PWD), PR interval, QRS duration, and corrected QT interval (QTc).Methods and findingsUncorrelated (r2 < 0.001) genome-wide significant (p < 5 × 10-8) single nucleotide polymorphisms (SNPs) were extracted from genome-wide association studies (GWAS) on body mass index (BMI, n = 806,834), waist:hip ratio adjusted for BMI (aWHR, n = 697,734), height (n = 709,594), weight (n = 360,116), fat mass (n = 354,224), and fat-free mass (n = 354,808). Genetic association estimates for the outcomes were extracted from GWAS on PR interval and QRS duration (n = 180,574), PWD (n = 44,456), and QTc (n = 84,630). Data source GWAS studies were performed between 2018 and 2022 in predominantly European ancestry individuals. Inverse-variance weighted MR was used for primary analysis; weighted median MR and MR-Egger were used as sensitivity analyses. Higher genetically predicted BMI was associated with longer PWD (β 5.58; 95%CI [3.66,7.50]; p = < 0.001), as was higher fat mass (β 6.62; 95%CI [4.63,8.62]; p < 0.001), fat-free mass (β 9.16; 95%CI [6.85,11.47]; p < 0.001) height (β 4.23; 95%CI [3.16, 5.31]; p < 0.001), and weight (β 8.08; 95%CI [6.19,9.96]; p < 0.001). Finally, genetically predicted BMI was associated with longer QTc (β 3.53; 95%CI [2.63,4.43]; p < 0.001), driven by both fat mass (β 3.65; 95%CI [2.73,4.57]; p < 0.001) and fat-free mass (β 2.08; 95%CI [0.85,3.31]; p = 0.001). Additionally, genetically predicted height (β 0.98; 95%CI [0.46,1.50]; p < 0.001), weight (β 3.45; 95%CI [2.54,4.36]; p < 0.001), and aWHR (β 1.92; 95%CI [0.87,2.97]; p = < 0.001) were all associated with longer QTc. The key limitation is that due to insufficient power, we were not able to explore whether a single anthropometric measure is the primary driver of the associations observed.ConclusionsThe results of this study support a causal role of BMI on multiple ECG indices that have previously been associated with atrial and ventricular arrhythmic risk. Importantly, the results identify a role of both fat mass, fat-free mass, and height in this association.

Dataset Information

Machine learning enables noninvasive prediction of atrial fibrillation driver location and acute pulmonary vein ablation success using the 12-lead ECG.

Background

Objectives

Methods

Results

Conclusion

Publications

Machine learning enables noninvasive prediction of atrial fibrillation driver location and acute pulmonary vein ablation success using the 12-lead ECG.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets