Early Detection of Breast Cancer Using Machine Learning: A Tool for Enhanced Clinical Decision Support

Abstract

Breast cancer arises when there is an abnormal increase in breast tissue, resulting in the creation of lumps or irregular cell layers.  This cancer ranks as the second most common among women worldwide, trailing only melanoma, and primarily impacts those over 50 years old, although it can manifest at any age.  Timely identification and robust preventive measures are essential for minimizing health risks associated with cancer.  Clinical trials in cancer prevention are persistently investigating innovative approaches for early diagnosis and treatment. This research utilizes machine learning methods to categorize breast cancer tumours as benign or malignant, facilitating prompt clinical decision-making.  The dataset utilized for this study was obtained from Kaggle and underwent preprocessing and exploratory data analysis, incorporating correlation matrices and Principal Component Analysis (PCA) for dimensionality reduction and data visualization. Four supervised machine learning algorithms were assessed: Decision Tree, Logistic Regression, Bagging, and Random Forest.  The evaluation of the models was conducted using metrics such as accuracy, precision, recall, F1-score, confusion matrix, and Area Under the Curve (AUC).  The Random Forest classifier demonstrated the highest accuracy at 98%, showcasing its exceptional ability to manage the provided dataset effectively.  Furthermore, the Bagging algorithm exhibited the highest AUC value at 99%, underscoring its effectiveness in differentiating between benign and malignant cases.  The results indicate that ensemble techniques, especially Random Forest and Bagging, serve as valuable instruments for breast cancer classification, potentially aiding clinicians in achieving early and precise diagnoses.