The early and accurate detection of Atrial Fibrillation (AFib) and Arrhythmia from Electrocardiogram (ECG) data is crucial to minimize the risks of heart-related complications. This paper introduces a Machine Learning (ML)-based solution that can detect and classify heart-related conditions using data collected from 12-lead ECG. The solution primarily focuses on detection and classification between Normal Sinus Rhythm (NSR), AFib, and Various Arrhythmias (VA) using time-domain and frequency-domain features. Five ML algorithms were evaluated, namely, Support Vector Machines (SVM), Random Forest (RF), Xtreme Gradient Boosting (XGBoost), K-Nearest Neighbors (KNN), and Adaptive Boosting (AdaBoost). The models used the PTB-XL dataset providing access to 12-lead ECG records for 6,428 patients. Results show that XGBoost outperforms the other models, achieving an accuracy of 85.67%. Time-domain features were found to be more reliable for short-duration ECG recording. The solution was deployed into an E-Hospital platform to provide diagnostic assistance for healthcare professionals.