The accurate categorization of breast tumors as either malignant or benign is crucial for effective diagnosis and treatment planning. This paper introduces a comprehensive evaluation addressing eight different machine learning models applied to the dataset used for breast cancer diagnosis in Wisconsin, which includes 569 instances featuring 30 attributes derived from digitized fine needle aspirate (FNA) biopsy samples. The algorithms assessed include Naive Bayes (NB), Logistic Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision Tree (DT), Random Forest (RF), AdaBoost, and XGBoost. Each model’s performance was assessed by means of several metrics such as accuracy, precision, recall, F1 score, specificity, confusion matrix, Matthews correlation coefficient, logarithmic loss, and area under the curve (AUC). Additionally, we explored ensemble techniques, including simple and weighted methods, alongside Voting Classifier Systems like Stacking and Boosting to enhance predictive accuracy. Our comparative analysis highlights the most effective strategies for early breast cancer detection and offers insights into developing AI-driven diagnostic tools. The study supports the integration of these models into web-based platforms, including the TIASM (TIASM: Téchnique de l’IA pour le Soutien de la Médecine) platform, emphasizing the critical role of model selection in medical decision support systems. These findings underscore the capabilities of machine learning to improve breast cancer prediction and patient outcomes.