Machine Learning for Arrhythmia Detection: Integrating ECG Biomarkers for Improved Diagnostic Performance

Cardiac arrhythmia detection is crucial for early diagnosis and treatment of cardiovascular diseases. Traditional electrocardiogram analysis relies on manual interpretation, which can be prone to human errors. In this study, artificial intelligence models, such as fine tree, support vector machine, and neural network were trained using tortuosity and RR interval as key features. The acquired data was obtained from 12-lead ECG recordings belonging to 14 control patients and 14 patients with arrhythmia. The models' performance was evaluated through specificity, sensitivity, accuracy and area under the curve metrics. Results showed that the SVM model achieves the highest accuracy (85.7%) and sensitivity (78.57%), demonstrating a better performance. These findings highlight AI's potential to enhance arrhythmia detection, improving efficiency and accuracy in diagnosis.