A Comparative Analysis of Machine and Deep Learning Models in the Early Detection of Breast Cancer

Breast cancer continues to be a major medical concern, being as the second leading cause for cancer-related fatalities in women. In 2017, over 250,000 new instances of breast cancer were identified in the United States, highlighting a global concern for medical researchers and healthcare professionals. Artificial Intelligence (AI) has played a pivotal role in the early detection of various diseases, including breast cancer. This paper conducts an in-depth comparative analysis of the most utilized machine learning and deep learning models for breast cancer detection. A total of 19 models, including SGD, SVM, NuSVC, LinearSVC, KNN, WKNN, GNB, RF, ExtraTree, Decision Tree, Dummy Classifier, Gradient Boosting, LightGBM, XGBoost, AdaBoost, CatBoost, LR, MLP, and ANN, were trained on the Wisconsin Diagnostic Breast Cancer (WDBC) dataset. These models were evaluated using standard metrics and subsequently compared. The comparative analysis reveals a range of effectiveness among the machine learning models. The AdaBoost Classifier was the best performer, achieving a testing accuracy of 97.37%, followed by LightGBM, Random Forest, Gradient Boosting, XGBoost, and CatBoost (CB), each showing accuracies above 95%. Furthermore, the AdaBoost Classifier attained the highest F1-score of 96.47%, indicating an optimal balance between precision and recall. The findings of this study underline the critical importance of selecting appropriate AI models to improve the accuracy of breast cancer diagnosis.