Detecting Anomaly in ECG Signal Using Time Series Forecast Approach

Detecting Anomaly in ECG signal using Time series forecast approach is a novel approach in the health care industry to improve the monitoring of the ECG signals. Machine learning algorithms are used to detect anomalies in the ECG signal. The ECG signals are divided into normal, non-beat and abnormal signals. A recurrent neural network (RNN)architecture, Long Short-term Memory (LSTM) can be used to develop a predictive model about the healthy and anomalous signals. LSTM works on raw signals. There is no need to preprocess the dataset before feeding it to the network when LSTM is being used, this is an added advantage to handling long-term dependencies. The proposed approach focuses on the abnormal ECG signals. Using time-series forecasting technique, a short-term forecasting of the ECG signals is performed. A recall of 99% is observed when LSTM model is used. On the forecasted ECG signal, anomaly detection is done using the machine learning algorithms. This will help the doctors to predict any anomaly that may occur soon.