Training machine learning models with synthetic data improves the prediction of ventricular origin in outflow tract ventricular arrhythmias

Abstract

In order to determine the site of origin (SOO) in outflow tract ventricular arrhythmias (OTVAs) before an ablation procedure, several algorithms based on manual identification of electrocardiogram (ECG) features, have been developed. However, the reported accuracy decreases when tested with different datasets. Machine learning algorithms can automatize the process and improve generalization, but their performance is hampered by the lack of large enough OTVA databases. We propose the use of detailed electrophysiological simulations of OTVAs to train a machine learning classification model to predict the ventricular origin of the SOO of ectopic beats. We generated a synthetic database of 12-lead ECGs (2,496 signals) by running multiple simulations from the most typical OTVA SOO in 16 patient-specific geometries. Two types of input data were considered in the classification, raw and feature ECG signals. From the simulated raw 12-lead ECG, we analyzed the contribution of each lead in the predictions, keeping the best ones for the training process. For feature-based analysis, we used entropy-based methods to rank the obtained features. A cross-validation process was included to evaluate the machine learning model. Following, two clinical OTVA databases from different hospitals, including ECGs from 365 patients, were used as test-sets to assess the generalization of the proposed approach. The results show that V2 was the best lead for classification. Prediction of the SOO in OTVA, using both raw signals or features for classification, presented high accuracy values (>0.96). Generalization of the network trained on simulated data was good for both patient datasets (accuracy of 0.86 and 0.84, respectively) and presented better values than using exclusively real ECGs for classification (accuracy of 0.84 and 0.76 for each dataset). The use of simulated ECG data for training machine learning-based classification algorithms is critical to obtain good SOO predictions in OTVA compared to real data alone. The fast implementation and generalization of the proposed methodology may contribute towards its application to a clinical routine.

Synthetic Force-Field Database for Training Machine Learning Models to Predict Mobility-Preserving Coarse-Grained Molecular-Simulation Potentials

“…In pursuit of developing methodologies to study physical systems, the scientific community has consistently relied on examining realistic models. To gain a deeper understanding of physical systems and advance methodologies through data-driven approaches, the data itself need not be confined to these physically realistic models. − This realization unveils a new avenue for data generation, as synthetic systems − are not bound by the same limitations as their physical counterparts. In this article, we demonstrate the generation of synthetic data and its subsequent utilization in studying physically relevant systems through the implementation of ML models.…”

Non-invasive Localization of the Ventricular Excitation Origin Without Patient-specific Geometries Using Deep Learning

“…We originally expected that transfer learning techniques would need to be applied. In a work published after the completion of out study, the authors also found simulated data to be at least beneficial for the performance on clinical data [12]. If clinical data was available in a larger quantity, as it was to the authors of [12], results could have further improved by using a part of the clinical data as training data, incorporated into the training process.…”

“…In a work published after the completion of out study, the authors also found simulated data to be at least beneficial for the performance on clinical data [12]. If clinical data was available in a larger quantity, as it was to the authors of [12], results could have further improved by using a part of the clinical data as training data, incorporated into the training process. Unfortunately at the current state, localization errors obtained from our implemented methods could be as high as 110 mm on clinical data.…”

“…Fully automatic approaches still relying on patient-specific geometries or usable for only one ventricle were successfully implemented in [51,67,65]. A very recent work classified the presence of ventricular outflow tract arrhythmia fully automatically using simulations to boost results on patient data [12].…”

ECG-based estimation of respiration-induced autonomic modulation of AV nodal conduction during atrial fibrillation

“…For example, synthetic images were generated to train neural networks to track cardiac motion and calculate cardiac strain ( Loecher et al, 2021 ), estimate tensors from free-breathing cardiac diffusion tensor imaging ( Weine et al, 2022 ), and predict end-diastole volume, end-systole volume, and ejection fraction ( Gheorghita et al, 2022 ). Furthermore, synthetic photoplethysmography (PPG) signals were generated to detect bradycardia and tachycardia ( Sološenko et al, 2022 ), and synthetic electrocardiogram (ECG) signals were generated to detect r-waves during different physical activities and atrial fibrillation ( Kaisti et al, 2023 ), and to predict the ventricular origin in outflow tract ventricular arrhythmias ( Doste et al, 2022 ).…”