Electrocardiogram Signal Forecasting using Iterated And Direct Method Based on Artificial Neural Network
Electrocardiogram (ECG) is the electric activity of heart. It is widely used for the identification and prognosis of cardiovascular disease. In this paper forecasting of electrocardiography signal is presented using two forecasting methods i.e. iterated method and direct method. These methods are based on Back propagation Neural Network (BPNN) Model. Comparison of these two methods has been presented, It is found that direct method outperform iterated method for forecasting of ECG signal. It is also found that mean square error (MSE) remains small in case of direct method for 3 steps ahead forecasting after that MSE increased rapidly. Clinical information in forecasted and actual signals is extracted by developing automatic ECG analyzer software. It is found that the clinical information was preserved in three steps ahead forecasting using direct method. It is concluded that neural networks have much potential for forecasting of ECG signals.