Women health issue: machine learning ensemble techniques to diagnosis breast cancer

A model for identifying and displaying important indicators of breast cancer (benign or malignant) is developed using machine learning methods in this study.

The major breast cancer prognostic factors were identified using support vector machines (SVM), K-nearest neighbor (K-NN), gaussian naive Bayes (GNB), random forest (RF), logistic regression (LR), extra tree classifier (ET), decision tree (DT), gradient boosting (GBC), and adaboost. The predictions are then transformed into a new data frame representing the state of breast cancer patients to perform advanced SVM modeling on the test set. It comprise of ten independent variables and one dependent variable, which alludes to the patient status (benign or malignant). The method’s efficacy is evaluated using a confusion matrix, ROC AUC curve, training score and CV score, precision-recall curve, and accuracy. Current research shows that these SVM classifiers perform better than ensemble classifiers and multiple machine learning classifiers in selecting the best model.

All algorithms had similar levels of model accuracy, with SVM having the highest level (97.85%) and decision trees having the lowest level (93.57%). In fact, the various machine learning algorithms tested in this study produced near-perfect results, suggesting that these techniques can be used as predictive tools for breast cancer detection. This is a fascinating discovery. The findings of this study could be used to develop tools for medical decision support.