Accurate Cardiac Arrhythmia Detection Using Horse Herd Optimization Algorithm-Enhanced Convolutional Neural Networks (CNN-HOA)

Abstract

Cardiac arrhythmias are significant contributors to global mortality, necessitating precise and timely diagnosis based on electrocardiogram (ECG) signals. This paper introduces an advanced diagnostic system utilizing a deep 1D Convolutional Neural Network (CNN) architecture integrated with the Horse Herd Optimization Algorithm (HOA) for automated feature extraction and classification. The primary objective is to overcome the limitations of manual hyperparameter tuning by using the HOA metaheuristic search capability to optimally tune critical parameters, specifically the Learning Rate and Batch Size. In the proposed framework, raw ECG signals from the standard MIT-BIH Arrhythmia Database are first preprocessed using the discrete wavelet transform with the "db6" mother wavelet to effectively suppress nonstationary noise and baseline wander. The denoised signals are then processed through a sequence of four convolutional blocks, featuring batch normalization and max-pooling, to extract hierarchical morphological features. Experimental results using 5-fold cross-validation demonstrate that the proposed CNN-HOA model achieves a state-of-the-art accuracy of 99.8%, with 99.9% sensitivity and specificity across five Association for the Advancement of Medical Instrumentation beat types: normal (N), supraventricular (S), ventricular (V), fusion (F), and unknown (Q). This study validates that the synergistic integration of modern swarm intelligence algorithms with deep learning provides a robust and highly efficient solution for high-precision medical signal processing and clinical-grade automated diagnosis.

Received: 23 October 2025 | Revised: 20 December 2025 | Accepted: 11 February 2026

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The ECG data used to support the findings of this study are available from the MIT-BIH Arrhythmia Database, which is publicly accessible via Figshare at https://doi.org/10.6084/m9.figshare. 845654.

Author Contribution Statement

Sajjad Mohammed Abdulkareem: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Data curation, Writing - original draft, Writing - review & editing, Visualization. Seyed Enayatallah Alavi: Conceptualization, Methodology, Resources, Writing - review & editing, Supervision, Project administration.