Breast cancer is the most commonly identified cancer among women and a major cause of death worldwide. Early detection significantly improves survival rates, but challenges remain in accurately differentiating between benign and malignant tumors. This study proposes a hybrid method combining machine learning algorithms, specifically Random Tree and J48, to improve diagnostic accuracy in breast cancer detection. The research used two datasets: the Ljubljana dataset, consisting of 286 patient samples with ten essential features such as tumor size, malignancy grade, and lymph node involvement, and the Wisconsin (WDBC) dataset, which contains 569 samples. Data preprocessing steps, including standardization, duplicate removal, and feature filtering, were applied to ensure data quality and relevance. The study compared the performance of multiple classifiers, including Naive Bayes, Support Vector Machines (SVM), and Random Tree, using metrics such as accuracy, precision, and recall. Results showed that the Random Tree algorithm achieved a remarkable accuracy of 97.9% on the Ljubljana dataset and 100% on the Wisconsin (WDBC) dataset, outperforming other single algorithms. Combined algorithms significantly increased accuracy. The hybrid method Random Tree + KNN and Random Tree + J48 achieved 96% and 99% accuracy, respectively, across the datasets. These combinations, with higher processing speed and improved accuracy, enhanced diagnostic efficiency and demonstrated greater suitability for clinical implementation and early breast cancer detection. This study highlights the potential of machine learning in advancing early breast cancer detection and sets a benchmark for scalable, accurate, and interpretable diagnostic tools in healthcare.