Enhanced Liver Cancer Risk Prediction using Hybrid Ensemble Machine Learning and Adaptive Synthetic Minority Oversampling

This paper presents a study of the life-threatening disease that affects millions of people globally. Liver disability is caused by various factors such as excessive alcohol consumption, viral infections, or some metabolic disorders, which can lead to serious threats like cirrhosis, liver failure, and eventually cancer. If treated at an early stage, it would be much more manageable and curable. In this work, we propose a machine learning (ML) model for identifying liver disease using ML techniques combined with SMOTE (Synthetic Minority Over-sampling Technique) to handle class imbalance. The methodology involves comprehensive data pre-processing steps such as missing value imputation, outlier removal, feature selection based on statistical significance, and balancing class distribution using SMOTE. Through extensive experimentation, our proposed ensemble model combining KNN and XGBoost demonstrates exceptional performance, achieving an outstanding AUC of 0.96 and an accuracy of 0.90. This approach not only outperforms alternative methods but also surpasses existing state-of-the-art techniques in liver disease diagnosis. We anticipate our model will significantly improve liver disease management and patient care by enabling clinicians to diagnose diseases earlier and more reliably, thereby providing a tool for better decision-making and outcomes.