Design and Implementation of an Automatic Arrhythmia Classification System Using a Hybrid Machine Learning Algorithm (Stacked Random Forest and J.48 Algorithms)


Authors : Onwuka, Ugochukwu C; Asagba, Prince O

Volume/Issue : Volume 5 - 2020, Issue 12 - December

Google Scholar : http://bitly.ws/9nMw

Scribd : https://bit.ly/2LgJJrU

Abstract : Arrhythmia is an abnormal condition of the heart that occurs when the electrical impulses that coordinate the heartbeats do not work properly, causing the heart to beat too fast, too slow, irregular or even have premature contractions; accurate and quick diagnosis can reduce fatalities, Thispaper focuses on design and development of a system that will automatically and accurately diagnose five super-class of arrhythmias recommended by theAssociation for the Advancement of Medical Instrumentation (AAMI) to be detectable by equipment/methodswhich includes: Normal(N), Supraventricular ectopicbeat (SVEB), Ventricular ectopic beat (VEB), Fusion beat (F) and Unknown beat (Q). To achieve this, the system uses a hybrid classification algorithm achieved by the combination of two better performing algorithm (Random Forest and J.48 ensembled with the stacking algorithm). The MIT-BIH ECG arrhythmia database accessible at Kaggle.com was used for training, testing and validation of the system. since this design is intended for edge devices or mobile devices, the design focuses on development of a system that uses less computation power, less application size and ability to give correct classification with less time. Our results showed that the designed system has an overall accuracy of 97.67%, average precision value of 0.977, average recall value of 0.977 and average F1- measure value of 0.977.

Keywords : Arrhythmia, Machine Learning, Random Forest, J.48, Stacking Ensemble.

Arrhythmia is an abnormal condition of the heart that occurs when the electrical impulses that coordinate the heartbeats do not work properly, causing the heart to beat too fast, too slow, irregular or even have premature contractions; accurate and quick diagnosis can reduce fatalities, Thispaper focuses on design and development of a system that will automatically and accurately diagnose five super-class of arrhythmias recommended by theAssociation for the Advancement of Medical Instrumentation (AAMI) to be detectable by equipment/methodswhich includes: Normal(N), Supraventricular ectopicbeat (SVEB), Ventricular ectopic beat (VEB), Fusion beat (F) and Unknown beat (Q). To achieve this, the system uses a hybrid classification algorithm achieved by the combination of two better performing algorithm (Random Forest and J.48 ensembled with the stacking algorithm). The MIT-BIH ECG arrhythmia database accessible at Kaggle.com was used for training, testing and validation of the system. since this design is intended for edge devices or mobile devices, the design focuses on development of a system that uses less computation power, less application size and ability to give correct classification with less time. Our results showed that the designed system has an overall accuracy of 97.67%, average precision value of 0.977, average recall value of 0.977 and average F1- measure value of 0.977.

Keywords : Arrhythmia, Machine Learning, Random Forest, J.48, Stacking Ensemble.

Never miss an update from Papermashup

Get notified about the latest tutorials and downloads.

Subscribe by Email

Get alerts directly into your inbox after each post and stay updated.
Subscribe
OR

Subscribe by RSS

Add our RSS to your feedreader to get regular updates from us.
Subscribe