Abstract: Electrocardiogram (ECG) is an intermittent sign, which mirrors the movement of the heart. From ECG a great deal data is gotten for typical and obsessive physiology of heart. The ECG signal is non-fixed in nature and extremely hard to dissect. Clinical perception takes long time and the sign is non-fixed. This paper presents a convolutional neural network algorithm for electrocardiogram signal classification. This system uses a ECG dataset which was downloaded from kaggle.com. The dataset signals were preprocessed to make sure that each segment conforms to a heartbeat. This dataset was read into the jupyter notebook by using pandas.read_csv function. The dataset was made into two, which are the training data and the testing data. After successful reading of the data signal from directory and solving of the in balance problem by means of data augmentation, the proposed model was trained using a convolutional neural network(CNN) algorithm with a total hidden layers of six, hiden size of 128, batch_size of 96, and number of epoch to be 10. After successful training, we had an accuracy of 99% at an epoch level of 10.

Keywords: Deep Learning, Electrocardiogram, ConvolutionalNeural Network, Heart Beat, Signal

PDF | DOI: 10.17148/IJARCCE.2021.10405

Open chat
Chat with IJARCCE