A Real-Time IoT Based Arrhythmia Classifier Using Convolutional Neural Networks

Abstract

With one in every four deaths that occur every year being due to a heart-related ailment, it is of utmost importance to study the symptoms, features, and cures for heart diseases so that timely action can be taken to detect and prevent fatalities. Arrhythmia is a type of heart ailment where the heart rate is irregular. It is caused due to erratic behavior of the electrical impulses that control the heartbeat. Detection and classification of arrhythmia are conventionally done manually by expert cardiologists through meticulous analysis of the electrocardiogram (ECG) waveform. Automatic and real-time ECG detection and analysis has gained importance in recent years especially due to the accelerating pace of advances in medical technologies. Therefore, the proposed system takes in data from an ECG sensor module (AD8232) and provides it as input to a trained Convolutional Neural Network (CNN) model in real-time. This model is capable of detecting various types of arrhythmias with accuracy greater than 90%. The classification results are then presented to the user through an interactive mobile application. A caretaker is also a part of the system, who is notified if in case the user's condition turns critical. Although a number of arrhythmia classification systems are implemented, the ease of access by the user and the interactiveness is highly limited. The implementation presented in this paper aims at providing an engaging user experience without compromising the performance and accuracy of measurements and predictions.