NARX-Wavelet Based Active Model for Removing Motion Artifacts from ECG

The continuous monitoring of the ECG (Electrocardiogram) signal has proved its potential for early detection of several life-threatening conditions in recent times. So the healthcare professional’s community is strongly emphasizing on developing low cost and efficient real-time ECG monitoring devices for continuous monitoring and predictive detection of life-threatening conditions which can be addressed at an early stage, before its lethal onset. In recent times it is observed that real-time ECG monitoring systems gained popularity due to their practical designs and comfortable attachments which do not interfere with patient/subject mobility as monitoring is performed in their natural environment, so due to motion artifacts or several other artifacts their outputs degrade at the analysis end. This paper addresses a serious issue of motion artifact filtering using a NARX-wavelet active model, which showed significant non-erroneous output as the filtering or de-noising is achieved through wavelet-based sampling rate independent technique and the motion artifacts are addressed by using NARX neural networks based on predictive noise cancellation in active mode.