Ensemble Classification System for Detection of Cardiovascular Diseases Using Electrocardiogram Signal

Early detection of Myocardial Infarction (MI) and Congestive Heart Failure (CHF) Cardiovascular Diseases (CVDs) are challenging diseases for cardiologist practioners to reduce the mortality rate.This paper deals with the design and development of an automated Ensemble Classification system using optimized Heterogeneous Features set viz. Morphological/Structural and Statistical Non Linear features of Electrocardiogram (ECG) Signal.ECG is non-invasive and vital clinical therapeutic agent deployed for taking intelligent health care prediction of MI and CHF.In this approach, the ensembles of classifiers are performed by taking into account diversity and accuracy of multi classifiers in intelligible hybridization manner with majority voting technique in ECG pattern recognition.Proposed methodology achieved the maximum Accuracy, Sensitivity, Specificity of 99.75%, 99.72%, 99.85% respectively, along with Precision, Recall and F1-Score statistical indices ranging from 0.9 to 1 value, taking into account 300 patient's ECG signals collected from diverse databases.The time required for execution of the system is 0.55 seconds.Computation time is reduced to greater extend with directly evaluation of the features from the ECG signal analyzed on the morphological and statistical domain, so, the detection of R-peaks are eliminated with the proper selection of derivative levels.