Online Heartbeat Classification Using Low Cost Algorithms

This work proposed a full system that combines different algorithms to perform online heartbeat detection that uses dedicated limited hardware. Some of the techniques that we have utilized, such as the Pan Tompkins QRS detection algorithm, have been extensively tested and used in different heartbeat classification systems. Other techniques, such as dynamic segmentation and Hjorth parameters, have been previously shown to work for offline classification and take less computational resources. The proposed model tests how the different techniques integrate and work with no previous information about the signal. It also verifies their accuracy by using the MIT-BIH Arrhythmia dataset and evaluates the execution time. Although it offers good accuracy and it is able to perform fast classification on a conventional laptop, it exceeds the execution time required for online classification.