Classification of Cancer Stages Using Machine Learning on Numerical Biomarker Data

Abstract

Cancer staging is a crucial aspect of determining appropriate treatment strategies and predicting patient outcomes. Traditional diagnostic methods rely heavily on invasive procedures and qualitative assessments, which can be time-consuming and prone to subjective errors. This research aims to address these limitations by leveraging machine learning (ML) techniques to classify cancer stages using numerical biomarker data, such as C-reactive protein (CRP), tumor mutation burden (TMB), and lactate dehydrogenase (LDH). By applying models like Random Forest, Support Vector Machines (SVM), Gradient Boosting, and Multi-Layer Perceptron (MLP), we aim to improve the accuracy of cancer stage classification. Feature selection through Recursive Feature Elimination (RFE) was performed to enhance model efficiency by identifying the most significant biomarkers. The results show that ML models can effectively predict cancer stages, with Random Forest achieving the highest accuracy at 85%. This method offers a non-invasive, rapid, and scalable alternative to conventional diagnostic approaches, potentially improving clinical decision-making and patient care.