Cardiovascular diseases, especially heart attacks, are the most common causes of death and mortality. This paper discusses the ability to predict and prevent a heart attack using machine learning, particularly in XGBoost algorithms, to discuss potential applications in early prediction and prevention of heart attacks. Therefore, we utilized a tenfold cross-validation process to train and validate an XGBoost model to both regression (XGBRegressor) and classification (XGBClassifier) tasks. The regression model did achieve an average absolute error, as computed to an interval of 0.279. The classification model worked at an excellent performance in accuracy, particularly that had an area under curve, AUC calculated with an interval of 0.997, undoubtedly testifying its ability toward making the right risk assessments toward this population. High predictions well represent XGBoost grasping complex risk factors associated toward the occurrences of heart stroke. These findings underscore how machine learning can transform care in health, enabling prevention at the right time. They can open up prospects for further improvements in management of cardiovascular health. Improvements could be made so that these predictive accuracies improve, allowing for meaningful application in reducing heart attack instances.