IoT-Based Atrial Fibrillation Screening System, Development, and Prospective Study

Atrial fibrillation (AF) is a progressive cardiovascular disease leading to heart failure characterized by irregular and often very rapid heart rhythm, so early detection and treatment thereof are desirable. An AF screening system that can be used in daily life needs to be developed. We propose an AF screening system based on Internet of Things (IoT) technology, comprising a wearable cardiac sensor. A smartphone app communicates with the wearable sensor and the cloud. In this system, a 30-beat R-R interval (RRI) sample (about 30 s) of a suspected patient is recorded by the wearable sensor, uploaded to the cloud via the app, and analyzed with an RRI classification model implemented in the cloud server. The output of the system is three severity levels of AF: 1) normal; 2) mildly abnormal (frequent premature beats); or 3) severely abnormal (AF). The RRI classification model implemented in the cloud server was developed using a machine learning (ML) technique. The advantages of the proposed system are 1) three-level severity estimation that is compatible with clinical practice, 2) short test time (about 30 s), and 3) accessible to patients undergoing catheter ablation. We prospectively validated the AF screening system under development with 200 records of 100 suspected patients. The accuracy, false negative rate, and overlooking rate were 88.8%, 4.92%, and 2.63%, respectively. In addition, the system was compared with the AF detection function of the Apple Watch. The IoT-based AF screening system would contribute to early treatment of AF since the system can be easily used at home.