Published Paper Details:

Cardiovascular disease (CVD) remains one of the leading causes of mortality globally, highlighting the need for early and accurate detection systems. This project presents a comprehensive approach for predicting cardiovascular disease using ensemble machine learning techniques. The system comprises an image processing module that pre processes ECG images and a machine learning module to analyze the extracted features. The image processing phase involves converting ECG images to grayscale, applying Gaussian smoothing to reduce noise, and contour detection for waveform extraction. The next phase integrates Principal Component Analysis (PCA) for dimensionality reduction, aimed at enhancing the model's efficiency by minimizing redundant data while retaining crucial features. A VotingClassifier ensemble method is then employed, combining multiple machine learning classifiers including Support Vector Machine (SVM), k Nearest Neighbors (kNN), Random Forest, Gaussian Naive Bayes, and Logistic Regression. The use of a soft voting strategy aggregates predictions based on probability scores, improving robustness and prediction accuracy. This modular architecture leverages the power of ensemble learning, utilizing ECG data to make reliable predictions. Initial results demonstrate the effectiveness of the proposed model in identifying patterns associated with cardiovascular disease. The final system is implemented using the scikit-learn and Streamlit libraries, providing an intuitive front-end for medical professionals to upload ECG images and view predictive results. This work underscores the potential of combining image processing and ensemble learning for non-invasive diagnosis of heart conditions.