Effective ECG classification using single layer neural network with data pre-processing

With appropriate linear pre-processing, such as key feature e t frac i ion o r discrete cosine transformation, a simp l e single-layer feedforward neural network can be trained t o classify ECG patterns on reduced dimension data s e f s with excellent accuracy. A comparison of muliiple oulpui single-layer feedforward neural networks demonstrates that these networks classify our 3 ECG paiierns wi fhin U classification sensi t iv i ty of ai least 0.5 with the appropriate pre-processed data.