Heart disease remains a leading cause of mortality, driven by factors like sedentary lifestyles, stress, and increasing rates of hypertension and diabetes. Early prediction is crucial, as timely diagnosis and intervention can significantly reduce the risk of fatal heart attacks. Leveraging machine learning for heart disease prediction enables healthcare providers to analyze patient data more accurately and efficiently, identifying high-risk individuals and promoting preventive care tailored to their specific health indicators. Aim to enhance heart disease prediction accuracy by evaluating the effectiveness of various machine learning algorithms and identifying key predictive features. We analyzed a dataset of patient health metrics, including age, cholesterol levels, chest pain type, and maximum heart rate, to determine the likelihood of heart disease. Among various models Random Forest and Decision Tree demonstrated the highest accuracy at 98.54%, followed closely by Gradient Boosting at 93.17%. Feature that are important for prediction are chest pain type, number of major vessels, maximum heart rate achieved, and ST depression were the most significant in predicting heart disease. The results underscore the effectiveness of ensemble methods in achieving high predictive accuracy and highlight critical health indicators, providing insights for healthcare diagnostics and risk assessment.