Machine Learning approach for TWA detection relying on ensemble data design

Abstract

Background and objective: T-wave alternans (TWA) is a fluctuation of the ST-T complex of the surface electrocardiogram (ECG) on an every-other-beat basis. It has been shown to be clinically helpful for sudden cardiac death stratification, though the lack of a gold standard to benchmark detection methods limits its application and impairs the development of alternative techniques. In this work, a novel approach based on machine learning for TWA detection is proposed. Additionally, a complete experimental setup is presented for TWA detection methods benchmarking.

Methods: The proposed experimental setup is based on the use of open-source databases to enable experiment replication and the use of real ECG signals with added TWA episodes. Also, intra-patient overfitting and class imbalance have been carefully avoided. The Spectral Method (SM), the Modified Moving Average Method (MMA), and the Time Domain Method (TM) are used to obtain input features to the Machine Learning (ML) algorithms, namely, K Nearest Neighbor, Decision Trees, Random Forest, Support Vector Machine and Multi-Layer Perceptron.

Results: There were not found large differences in the performance of the different ML algorithms. Decision Trees showed the best overall performance (accuracy $0.88\pm 0.04$ , precision $0.89\pm 0.05$ , Recall $0.90\pm 0.05$ , F1 score $0.89\pm 0.03$ ). Compared to the SM (accuracy 0.79, precision 0.93, Recall 0.64, F1 score 0.76) there was an improvement in every metric except for the precision.

Conclusions: In this work, a realistic database to test the presence of TWA using ML algorithms was assembled. The ML algorithms overall outperformed the SM used as a gold standard. Learning from data to identify alternans elicits a substantial detection growth at the expense of a small increment of the false alarm.

Keywords: Cross Validation (CV); Electrocardiogram (ECG); Machine Learning (ML); Modified Moving Average Method (MMA); Repolarization; Spectral Method (SM); Time Method (TM); T–Wave Alternans (TWA).

Figure 1

Flow chart of the database assembling.

Figure 2

Input vectors $\mathbf{x}={[x_1,x_2,x_3]}^T$ for the ML algorithms. The normalized values from the TM, SM and MMA are represented in the Z, Y and X axis respectively.

Figure 3

Cross Validation scheme of a model, where PG stands for Patient Group, k_j refers to the hyperparameters tuple of a model, and ϕ stands for the metric utilized for the optimization problem. Patient group test set (PG₆) is fitted with optimum hyperparameters obtained from CV. Notice that each PG corresponds to a disjoint group of patients.

Figure 4

Diagram representation of permutation scheme, where PG stands for Patient Group. Notice that each PG corresponds to a disjoint group of patients.

Figure 5

Flow chart of ML setup followed in this work.

Figure 6

Classification performance for different ML algorithms in terms of mean and standard deviation. Every pair of bars stands for the metric value in the train (left) and test (right) sets.