This analysis aims to design a diagnostic tool using a predictive model for the early detection of breast tumor malignancy by applying Python programming software and Machine Learning. We employed supervised and unsupervised methods, including PCA, ISOMAP, and clustering, to perform a preliminary analysis and explore the data structure. Subsequently, supervised methods are applied to train different models, using histopathological data and pre-accuracy metrics. The results highlight that LDA offers an optimal balance between accuracy and recall, being suitable for accurate differentiation between benign and malignant tumors. In response to these findings, a graphical interface trained with LDA was developed, aimed at improving early identification of tumor malignancy. This tool promises to facilitate more accurate prognostication and therapeutic intervention for patients.