Random Forest Classifier Based ECG Arrhythmia Classification

Random Forest Classifier Based ECG Arrhythmia Classification

V.Mahesh (V.MaheshPSG College of Technology, India), A. Kandaswamy (PSG College of Technology, India), C. Vimal (PSG College of Technology, India) and B. Sathish (PSG College of Technology, India)
DOI: 10.4018/978-1-4666-1755-1.ch013


Heart Rate Variability (HRV) analysis is a non-invasive tool for assessing the autonomic nervous system and for arrhythmia detection and classification. This paper presents a Random Forest classifier based diagnostic system for detecting cardiac arrhythmias using ECG data. The authors use features extracted from ECG signals using HRV analysis and DWT for classification. The experimental results indicate that a prediction accuracy of more than 98% can be obtained using the proposed method. This system can be further improved and fine-tuned for practical applications.
Chapter Preview


ECG data for the analysis and classification were obtained from the MIT-BIH arrhythmia database and signals from ML-II leads are used. This digital data has been sampled at a frequency of 360Hz and preprocessed to remove noise due to power-line interference, muscle tremors, spikes etc. The work presented in this paper focuses on several important arrhythmia types such as Paced beat (P), Atrial premature beat (A), Right bundle branch block beat (R), Left bundle branch block beat (L), Ventricular escape beat (E), Ventricular flutter wave (!), premature ventricular contraction (V), Fusion of ventricular and normal beat (F), Fusion of paced (f), Blocked Atrial Premature Beat (x) and the Normal beat segment (Normal). One minute segments of each beat type are extracted from these records for analysis. The number of segments of each type extracted from the database records is given in Table 1.

Table 1.
Number of segments of each type extracted from the database records
Type of ArrhythmiaNo of segments Extracted

Complete Chapter List

Search this Book: