ECG signal pattern recognition using grey relational analysis

The paper proposes a learning algorithm of grey relational analysis, called grey relational algorithm, to reduce the computational procedure of original method, and then applies this learning algorithm to electrocardiogram (ECG) signal pattern recognition. The training patterns for the learning algorithm are chosen several ECG signals from an ECG waveform database with patient diagnosis information. During the learning phase, each iteration only takes a training pattern into consideration to determine the influential factors of grey relational algorithm. After the learning process is finished, we can guarantee that the learning results are identical to the analytic results obtained by grey relational analysis. The similarity measure of other comparative ECG beats can be determined by the proposed method, and be furthermore identified as the diagnosis for the examined ECG. Simulation results show that the proposed method could perform not only a well computational efficiency but also a good classified accuracy.