Input Feature Selection in ECG Signal Data Modelling using Long Short Term Memory

One of the diseases that are a significant burden worldwide is cardiovascular disorders, diseases related to the work of the heart have a high probability of causing death. So we need a tool or model to detect the patient's heart signal against the risk of cardiovascular disorders. Electrocardiogram (ECG) recordings are often used to capture the propagation or propagation of electrical signals in the heart from the patient's body surface. Reading the ECG signal data is very tiring because every second, there are around 180 points that are captured which consist of the patient's pulse, movement, and breath. In this research, input feature selection will be carried out using the Long Short Term Memory method for ECG signal data. The results of the prediction of the ECG signal can be used to predict and treat cardiovascular disorders. Furthermore, the results of the model performance that the Long Short Term Memory model with one input, namely (t-1), is the best compared to using two or four input features.