COMPARATIVE ANALYSIS OF MACHINE LEARNING TECHNIQUES FOR ACCURATE LUNG CANCER PREDICTION

Abstract

Lung cancer is a major global health concern, being one of the most common and fatal cancers. Accurate early detection and prediction of lung cancer are crucial for improving patient outcomes, and machine learning (ML) algorithms offer promising solutions for enhancing diagnostic accuracy. This study evaluates the performance of five ML algorithms—XGBoost, LightGBM, AdaBoost, Logistic Regression, and Support Vector Machines (SVM)—for lung cancer prediction. Utilizing a diverse dataset with attributes such as demographic variables, lifestyle factors, clinical features, and environmental exposures, we conducted a comprehensive analysis involving data preprocessing, feature selection, and model training. Our results indicate that XGBoost achieved the highest performance across all metrics, including accuracy (97.50%), sensitivity (96.80%), specificity (98.00%), and F-1 score (97.50%). LightGBM also performed well but slightly lagged behind XGBoost. AdaBoost, Logistic Regression, and SVM exhibited lower performance compared to the top two models. The correlation analysis revealed significant predictors of lung cancer, such as smoking history, air pollution, and family history. This study underscores the superiority of XGBoost in lung cancer prediction and suggests that future work should focus on expanding datasets, refining feature engineering, and integrating ML models into clinical practice for enhanced diagnostic capabilities.