A Comparative Analysis of Metaheuristic Algorithms on Hybrid Features for Feature Selection on ECG-Based Arrhythmia Detection

Electrocardiogram (ECG)-based arrhythmia detection is crucial for diagnosing cardiovascular diseases promptly. However, the high-dimensional nature of ECG signals poses challenges for accurate detection. Feature selection plays a vital role in enhancing classification performance by identifying relevant features. This paper presents a comparative analysis of metaheuristic algorithms for feature selection in ECG-based arrhythmia detection using a hybrid feature set based on morphological and heart variability features. We evaluate the performance of Grey Wolf Optimization (GWO), and Firefly Algorithm (FFA) in terms of classification accuracy, sensitivity, Positive precision, and specificity. The experimental results shows that GWO performs FFA better in selecting relevant features for arrhythmia classification by 2.8%. However, both algorithms demonstrate the efficacy of metaheuristic algorithms and hybrid features in improving arrhythmia detection and provide insights into their comparative performance.