Prediction of Pancreatic Cancer Through Biomarkers Using Machine Learning Techniques: An Approach for Early Diagnosis

Abstract

This article explores the prediction of pancreatic cancer using CA 19-9 and CA 125 biomarkers with three machine learning models: Gradient Boosting, Random Forest, and Logistic Regression. The study evaluates their effectiveness through 10-fold cross-validation. Results showed competitive performance, with the Logistic Regression model displaying the highest accuracy, precision, and F1-score, indicating its potential for early diagnosis. Integrating biomarkers and machine learning promises for improving pancreatic cancer prediction and patient outcomes.