ECG Data Classification With Privacy Preservation in the IoT Context

Myocardial infarction (MI) can go undetected and leads to permanent heart damage or death. Its early detection with the proper medication can save lives. These days, continuous monitoring of the heart is possible due to IoT devices. Various cardiovascular diseases can be identified without the expertise of a cardiologist, only by analyzing the ECG signals. Most of the time, the inputs provided to an AI model are sensitive data. In order to avoid sending patients private data through the network and be exposed to multiple attacks, Federated Learning (FL) is implemented in the myocardial infarction detection system. The ECG signals are classified by a one-dimensional convolutional neural network (1D-CNN) which efficiently differentiates the subjects with MI from those who are healthy.