An intelligent human‐centric systems to diagnose breast cancer using machine learning and optimized feature selection techniques

Abstract

All over the world, breast cancer (BC) is the leading cause of cancer mortality among women. Computer‐aided methods can assist in early diagnosis. The proposed approach used SMOTE filter with Ch² test techniques for class balance and feature section using eight different ML models Gaussian Naive Bayes (GNB), Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM) with Linear and Radial Basis Function (RBF), Logistic Regression (LR), K‐nearest neighbor (KNN) and eXtreme Gradient Boosting (XGBoost). A Ch² test determines the top five features—glucose, HOMA, resistin, BMI, and insulin. Metrics such as accuracy, precision, recall, and F1‐Score are used to compare the performance of models. More than 99% accuracy was achieved by the proposed XGBoost model. Compared to the other breast cancer prediction models, the proposed model had an average accuracy improvement of 9.30%. As a result of our proposed model, breast cancer diagnosis will be more efficient based on risk factors. The proposed prediction model can also predict various breast cancer features. In addition to improving diagnostic decision‐support systems, the proposed model should be able to predict breast cancer disease accurately.