A simplified Approach for Accurate Arrythmia Detection using Automated Machine Learning

This paper introduces a straightforward and computationally inexpensive approach to classifying heartbeat anomalies based on ECG signals. Cardiac arrests are often associated with irregular heart rhythm, making arrythmia a major factor to cardiac events when not sufficiently monitored. In this context, the current paper presents a comprehensive analysis and arrythmia prediction for 44 patients. For each patient in the study, heartrate signals manually labeled by physicians were available. The preprocessed time-series were used to train machine learning models available in various automated frameworks, allowing for accurate binary classification. The highest scores after benchmarking were obtained by LightGBM. Our proposed method provides similar results in terms of classification performance with state-of-the-art algorithms for arrythmia detection. The current work thus introduces a simplified pipeline, improvement in prediction time and classification accuracy.