A major global health concern that primarily affects children, congenital heart diseases (CHDs) highlight the urgent need for prompt and precise diagnostic measures. Using medical imaging datasets, this study looks into how machine learning methods can be used to enhance CHD diagnosis and prognosis. Advanced feature extraction methods, including the Gray Level Co-occurrence Matrix (GLCM), Histogram of Orientated Gradients (HOG), and Local Binary Patterns (LBP), were successfully used to extract significant insights from heart images. After analyzing a number of machine learning models, including SVM, Random Forest, XGBoost, and Decision Trees, XGBoost was shown to be the best performer, outperforming the others in terms of accuracy, precision, recall, and F1 score. The findings demonstrate the revolutionary potential of machine learning, especially XGBoost, in facilitating precise and early CHD identification. This research contributes to the advancement of pediatric precision medicine by providing scalable and non-invasive diagnostic tools, empowering clinicians to enhance patient care and optimize treatment plans. By combining robust data preprocessing techniques with powerful algorithms, this study sets the stage for a proactive and technology-driven approach to CHD management.