A Machine Learning Approach For Atrial Fibrillation Detection From Simulated ECG Signals

Atrial fibrillation (AFib) is one of the most prevalent cardiac arrhythmias and a leading cause of stroke and hospitalization in elderly patients. Timely diagnosis is crucial, yet continuous ECG monitoring and expert evaluation remain difficult to scale outside clinical environments. This work presents a simple yet effective machine learning model capable of detecting AFib using synthetic ECG data. A dataset of 1000 samples was simulated, evenly split between normal rhythm and AFib. Three key features were extracted: RR interval variability, heart rate variability (HRV), and a simulated index for P-wave presence. A Random Forest classifier trained on this dataset achieved an accuracy of 99.7%, precision of 100%, recall of 99.3%, and an area under the ROC curve (AUC) of 1.0. These results suggest that machine learning techniques, even when applied to simplified data, can offer valuable tools for early arrhythmia screening, potentially enabling low-cost, scalable solutions for wearable health monitoring