Machine Learning Techniques for Breast Cancer Prediction: A Comprehensive Review on Techniques and Datasets
Archana Singh · Communications on Applied Nonlinear Analysis · 2025
Breast cancer (BC) is a key health concern worldwide; early detection and accurate diagnosis are crucial for improving patient outcomes. Machine learning techniques have shown promise in revolutionizing breast cancer diagnosis. This review covers various machine learning techniques, ranging from classic algorithms like decision trees and k-nearest neighbor (KNN) to advanced methodologies such as ensemble learning and deep learning. The variation in accuracy metrics and the lack of standardized evaluation methodologies make it challenging to directly compare the performance of diverse algorithms. This study discusses the data sources and methodology employed in the examined studies, as well as comparing various machine learning approaches. The findings of this work indicate that machine learning approaches may greatly enhance the diagnosis of breast cancer. The comparison analysis clarified that ensemble learning provided better results on the Wisconsin breast cancer dataset (WDBC), attaining the highest metrics with an accuracy, precision, recall, and an F1-score. Furthermore, the optimized framework demonstrated highest accuracy on ultrasound image data, underscoring its efficacy and robustness in medical diagnostics. This review provides a unique and critical analysis of the machine learning techniques and data sources used in breast cancer diagnosis and highlights the need for further research in this area.