ECG Data Analysis with IoT and Machine Learning

Regularity and irregularity in heartbeat rhythms are widely diagnosed through the analysis of electrocardiogram (ECG) recordings of the heart. However, clinical analysis comes with financial and scheduling costs that may not be necessary in the absence of heart illnesses. The goal of this research is to investigate if we can design a mobile health system that allows users to monitor and analyze their heart rhythms by using Internet of Things (IoT) techniques and machine learning methods. We proposed to have an IoT-based system embedded with a trained machine learning model to address this issue. The proposed system contains a few parts. An AD8232 sensor was used to collect ECG signals, which were then sent to a minicomputer, Raspberry Pi, through an analog-to-digital (ADC) converter. The digital ECG signals were fed to a Flask application and then sent to the machine learning model for classification. The machine learning model was trained on three different heart rhythms from the PhysioNet, including Normal Sinus Rhythm, Congestive Heart Failure, and Atrial Fibrillation. The training and testing processes were also in the IoT device. To support interaction for the public, a mobile web application was created to show the result of the classification as well as presenting real-time ECG data as a continuous graph. The result of the work demonstrates that IoT techniques not only allowed users to monitor their heartbeats, but it could also analyze real-time ECG signals with a trained machine learning model for anomaly detection.