mobilemodem / ecg-heartbeat-classification-deep-learning Goto Github PK

This project forked from ma7modhamed/ecg-heartbeat-classification-deep-learning

In this project, convolutional neural networks are proposed to classify sixteen different ECG beat types in the MIT-BIH arrhythmia database. We aimed to implement a computer-aided diagnosis system in intensive care to help doctors instead of keeping an eye on the patient 24 hours, the system automatically classifies heartbeats and alarm the doctor. We have tried Long short-term memory (LSTM), Visual Geometry Group (VGG16), Convolution Neural Network (CNN) and Inception. We have two architectures for classification, one stage and two stages. We tried more than one trail and we got best result from CNN. One Stage CNN has 91.17 % as average accuracy And 99.08 % as overall accuracy. Two stages CNN has 91.98 % as average accuracy And 98.74 % as overall accuracy.