Exploring machine learning approaches for breast cancer detection: a comparative perspective

Breast cancer, on the other hand, is the leading cause of death among women worldwide, necessitating early and timely diagnosis to reduce survival mortality rates. This traditionally relies on radiology tests such as mammography and, in some cases, biopsy, which have several clinical limitations in terms of invasiveness and a risk of false negatives/positives associated with the loss of precious time. This paper examines using machine learning techniques in improving diagnosis relating to breast cancer and reducing diagnostic errors. Using the Breast Cancer Wisconsin dataset, several ML models including Decision Trees, K-Nearest Neighbors, Support Vector Machines, Logistic Regression, Naive Bayes, and Neural Networks were evaluated. Performance metrics considered were accuracy, precision, recall, F1-score, and Area Under the Curve. The best neural network was MLP, outperforming other models and further showing how ML might offer accurate, non-invasive diagnostic alternatives. Further studies should be done to enable better real-world applications and clinical utility of such models.