Breast cancer prediction using machine learning classification algorithms

Abstract

In bioinformatics, the integration of machine learning has revolutionized disease diagnosis. Machine learning algorithms remove human limitations, offering more accuracy in diagnosing diseases like cancer. Breast cancer, the second most diagnosed cancer in women, often relies on mammography, which is only 70 % accurate, leading to potential misdiagnosis. Biopsies, though more reliable, are subject to human error and conflicting specialist opinions, often requiring multiple biopsies. The shortage of pathologists further complicates accurate and timely diagnoses. Machine learning can reduce these errors, providing faster and more precise results. In this study, a breast cancer dataset with 11 features is analyzed using eight machine learning classifiers. Results showed that Logistic Regression achieved the highest testing accuracy of 91.67 % without feature selection. After applying feature selection, classifiers like LGBM improved, with a notable 90.74 % accuracy. This study highlights the importance of integrating machine learning into healthcare, not only for breast cancer but for other diseases like heart disease and diabetes. Continued exploration and application of machine learning in bioinformatics will enhance its accessibility and effectiveness for medical professionals worldwide, leading to improved patient outcomes.

Topics & Concepts