Kernel modeling on Gaussian process to eliminate power line interference for ECG signal

An electrocardiogram (ECG) signal noise caused by different source e.g. power line interference (PLI), muscle and motion artifacts etc. Eliminating signal noise from ECG signal will increase the accuracy in diagnosis disease. In this paper, we propose sinusoid kernel function on Gaussian process (GP) to eliminate power line interference (50Hz or 60Hz) for electrocardiogram (ECG) signal. Expressing mathematic model to remove signal noise from ECG and compare the simulation result with state space recursive least squares (SSRLS). So we found that proposed algorithm has not transient time before the steady state that is the difference important point between GP algorithm and SSRLS algorithm. Furthermore, the mean square error (MSE) provides the different value noticeably.