Detecting Premature Ventricular Contraction by Using Regulated Discriminant Analysis with Very Sparse Training Data

Abstract

Pathological electrocardiogram is often used to diagnose abnormal cardiac disorders where accurate classification of the cardiac beat types is crucial for timely diagnosis of dangerous conditions. However, accurate, timely, and precise detection of arrhythmia-types like premature ventricular contraction is very challenging as these signals are multiform, i.e. a reliable detection of these requires expert annotations. In this paper, a multivariate statistical classifier that is able to detect premature ventricular contraction beats is presented. This novel classifier can be trained with a very sparse amount of expert annotated data. To enable this, the dimensionality of the feature vector is kept very low, it uses strong designed features, and it uses a regularization mechanism. This approach is compared to other classifiers by using the MIT-BIH arrhythmia database. It has been found that the average accuracy, specificity and sensitivity are above 96 percent, which is superior given the sparse amount of training data.

Searching for Premature Ventricular Contraction from Electrocardiogram by Using One-Dimensional Convolutional Neural Network

“…Considering that labeled ECG data is rare and precious, Lynggaarda suggested a multivariate statistical classifier that used robust designed features and a regularization mechanism. Even though this classifier's input is a very sparse amount of expert annotated ECG data, this model's average accuracy, specificity, and sensitivity are above 96% by using the MIT-BIH arrhythmia database [12]. Sokolova et al recommended a set of weighted shape parameters from the different QRS shape metrics and designed a two-stage PVC detection rule.…”