A supervised classification phenotyping approach using machine learning for patients diagnosed with primary breast cancer

This paper presents a methodology for the early detection and diagnosis of breast cancer using the Wisconsin dataset. The methodology involves four main steps, including data collection, preprocessing, feature selection, and classification. Fine needle aspiration technique is used to extract the ultrasound image features of breast cancer, and preprocessing is performed to eliminate outliers, null values, and noise. Redundant parameters are removed during the feature selection process to improve accuracy. Six machine learning algorithms, including Logistic Regression, Support Vector Machine, K-Nearest Neighbor, Random Forest, Decision Tree, and Gaussian Naive Bayes, are employed for the classification of the breast cancer dataset. Support Vector Machine and K-Nearest Neighbor achieved the highest accuracy, with Logistic Regression, Gaussian Naive Bayes, Random Forest, and Decision Tree having lower accuracy scores. The proposed methodology could aid in the timely detection and diagnosis of breast cancer, and help doctors in selecting the optimal clinical treatment plan for their patients. Further work will be carried out to investigate the effectiveness of additional preprocessing algorithms in improving the classification accuracy of the breast cancer dataset.