Analysis of Experimental Dataset for Potential Application in Non-Invasive Cancer Diagnosis

Computing and technology are essential to biomedical data analysis in modern medicine. This work analyses a biomedical dataset from a unique cancer diagnosis strategy using analytics and machine learning. The paper briefly describes the research background, data preprocessing methods, model, results, and conclusions. Feature importance was analysed using random forests, while control and diagnostic group differences were analysed using k-Nearest Neighbor, Random Forest, Support Vector Machine, and Logistic Regression. The study identified the most essential properties that the models use in training to distinguish between the classes in binary classification, which were evaluated by medical specialists. The proposed solution sets a baseline for using these data for diagnostics after machine learning models prove the proposed medical hypothesis and patterns.