Enhancing Prediction Precision and Reliability in Cervical, Lung, and Breast Cancer Diagnosis

Abstract

This study covers a wide range of cancers, focusing on cervical, lung, and breast cancer. Developing fast, accurate, and interpretable machine learning models for early diagnosis is critical to reducing the multifactorial mortality associated with these cancer types. Using a two-stage hybrid feature selection method, this study evaluates classification models using specific cervical, lung, and breast cancer data obtained from the UCI Machine Learning Repository. The cervical cancer dataset contains 36 features, the lung cancer dataset contains 16 features, and the breast cancer dataset contains 31 features. In the first stage, a random forest architecture is used for feature selection to identify features 5,7, and 7 that show a strong correlation with their cancer while reducing the difference between them. In Stage 2. Logistic regression (LR), naive bayes (NB), support vector machine (SVM), random forest (RF), and decision making (DT) were used to identify cervical cancer, lung, and breast cancer patients for five selections.