A Novel ECG Signal Quality Index Method Based on Skewness-MODWT Analysis and Detection of Arrhythmia with BPM

Because of their ease of use and compact size, wearable devices have become increasingly popular for continuous electrocardiogram (ECG) monitoring. It is necessary to do continuous, day-to-day, round-the-clock monitoring in order to identify anomalous symptoms that may be early warning indicators of serious illnesses. To keep the system running and increase the amount of time that devices can be used, cost-saving measures must come first. Nevertheless, not all recorded data are valuable due to many factors influencing the signal quality, including noise from muscles and motion aberrations. It is necessary to categorize and transmit just the useful signals in order to preserve wearable device batteries and server-side (cloud-based) processing costs related to processing and storage resources.

In order to assess signal quality and satisfy the aforementioned criteria, Signal Quality Indices (SQIs) have been created and studied. This paper presents a new SQI method for signal classification. Three things are added by the suggested method: Three methods are used to classify ECG signals: 1) use the lightweight exponentially weighted mean-variance (EWMV) equation to identify peaks; 2) define peaks that have the same shape as R-peaks by applying an adaptive threshold; and 3) use the maximal overlap discrete wavelet transform (MODWT) to classify signals into a new class that may contain signals relevant to pathological analysis. Our approach achieves the best sensitivity for both noisy and noiseless data sets, as shown by experimental findings. And finally with the use of extracted peaks in the previous stage, we can calculate Beats Per Minute (BPM) to determine the final health condition of heart i.e., Arrhythmia.

Keywords: Electrocardiogram (ECG), exponentially weighted mean-variance (EWMV), maximal overlap discrete wavelet transform (MODWT), signal quality indices (SQIs), BPM.