Detection of Cardiovascular Disease from ECG Image Using Optimized Weighted Average Ensemble Machine Learning Technique

cardiovascular diseases (CVD) stand as a prominent global cause of death in worldwide. Timely identification and intervention are paramount in dealing with heart disease, as its failure to detect and treated promptly can lead to life-threatening outcomes. The Electrocardiogram (ECG) Signals are mainly used as diagnostic tools that routinely assess the electrical and muscular function of the heart. Until recently, ECG records were predominantly maintained in paper form; thus, manually examining and re-examining the ECG paper records are often proved time-consuming and laborious tasks. Therefore, this research work utilizes, the power of machine learning and deep learning in predicting four different cardiac abnormalities named Abnormal Heartbeats (AHB), History of Myocardial Infarction (HMI), Myocardial Infarction (MI), and Normal Heartbeats (NHB) from paper-based ECG images. Different image pre-processing methods have been employed to improve the quality of individual ECG images and a novel lightweight deep neural model is proposed for extracting prominent features from each ECG images. Finally, we used an optimized weighted average ensemble technique for the final prediction. The grid search technique is used to find the optimal parameters and weights of each machine-learning model. Our proposed model shows state-of-the-art results in accuracy, precision, recall, and f1-score of 99.29%, outperforming the existing work. The model has attained an Area Under Curve (AUC) score of 99.96%.