An Efficient Analysis Scheme for Intelligent ECG Monitoring Devices

In the last few decades, consumer electronics in the field of biomedical devices has sought significant growth. With the advancement in artificial intelligence (AI), the biomedical devices such as electrocardiography (ECG) monitoring systems are now capable of performing automated analysis. However, there is significant room for enhancement in the efficiency of the available devices. This paper presents an efficient methodology for automated recognition of ECG signals. Here, the double density complex wavelet transform (DDCWT) method is employed for capturing the time-frequency (TF) information from the ECG signals. Different features from the output coefficients are captured and concatenated with the heart-rate variability information between ECG signals. This resulting vector carry sufficient information of each heartbeat and is classified using twin support vector machine (TSVM) scheme to classify into five categories. The classifier metrics are chosen by employing artificial bee colony (ABC) algorithm to enhance its performance. The experiments are conducted on the MIT-BIH data under subject oriented scheme where an accuracy of 97.20% is reported. The microcontroller-based implementation of the proposed methodology will result in the development an efficient monitoring system for consumers targeted for mass market.