Photoplethysmography Based Arrhythmia Detection and Classification

Arrhythmia is the most common cardiovascular disease caused due to abnormal heartbeat i.e. the heart may beat too slow, too fast. Many a time's irregular heartbeats may lead to heart attack, organ failure or even can cause death. Therefore, it becomes essential to detect the presence of arrhythmia at earliest. Electrocardiogram (ECG) and Photoplethysmograph (PPG) based sensors can be used for measuring the activity of the heart. However, both techniques are not providing enough information for the current detection of arrhythmia. To overcome these limitations in this paper, we present PPG based method that can be used for the detection of abnormality of heart. Firstly, signals preprocessed, then abnormalities are detected from the signals features and finally, classification is performed using different machine learning algorithms. PhysioNet database namely MIMIC II has been used for the evaluation of the proposed method. These databases are publically available following the standards developed by the Association for the Advancement of Medical Instrumentation (AAMI). Results show that SVM gives better accuracy (97.674%) compared to the other algorithms for the detection of arrhythmia pulses.