Heart disease is one of the leading causes of death worldwide, and therefore, accurate early detection methods are needed to help reduce mortality rates. One approach that can be applied is machine learning using classification techniques based on ensemble boosting algorithms. This study aims to compare the performance of two ensemble algorithms, namely Adaptive Boosting (AdaBoost) and Categorical Boosting (CatBoost), in classifying heart attack disease. The labels used in this study are positive and negative. The evaluation process was conducted using two testing techniques: percentage split with a ratio of 80% training data and 20% testing data, and 10-fold cross-validation. Model performance was evaluated based on accuracy, precision, and recall to comprehensively measure classification capability. The results show that in the percentage split method, CatBoost achieved the highest accuracy of 98.88%, while in k-fold cross-validation it reached 98.43%. Nevertheless, AdaBoost also demonstrated good performance, with all evaluation metrics exceeding 90%. Therefore, the best-performing model in this study is CatBoost with the k-fold cross-validation technique on the heart attack dataset.