Enhancing Breast Cancer Prediction Through Machine Learning Techniques

Breast cancer is the most frequent female cancer due to a variety of clinical, lifestyle, social, and economic variables. Machine learning methods are particularly effective in predicting breast cancer by uncovering patterns from a dataset. The goal of this study was to use several machine learning algorithms in predicting breast cancer using demographic, laboratory, and mammography data. An analytical investigation was conducted utilizing a database from the Motamed Cancer Institute (ACECR) in Tehran, Iran, consisting of 5178 independent records, 25% of which were biopsy records from breast cancer patients. Each record included 24 properties. Several machine learning techniques were utilized, including Random Forest (RF), XGBoost, Adaboost, and Stacking. The initial model was created using demographic and laboratory data (20 features), and the effect of all demographic, laboratory, and mammographic characteristics (24 features) in the model was later assessed. Deep learning based on random forest modelling outperformed the dough primitive modelling approach in chest radiographer assisted prediction, with 99.89% accuracy, 98.56% precision, 97.23% recall, and 0.02% FPR. Adoption of breast cancer diagnosis models that take into account numerous risk variables has the potential to improve early identification and management.