Ensemble Machine Learning Techniques for the Identification of Liver Diseases: A Comparative Approach

Liver diseases represent a significant global health issue, markedly impacting morbidity and mortality rates. Timely and precise diagnosis is essential for enhancing patient outcomes; however, conventional diagnostic methods frequently depend on invasive techniques, subjective assessments, and restricted datasets, leading to potential delays or inaccuracies in diagnosis. Advanced computational intelligence techniques like machine learning (ML) have already proven their efficacy and provide significant solutions for decision-making in medical diagnosis. Our objective is a comparative approach of stand-alone ML models, which includes support vector machine (SVM), random forest (RF), decision tree, and eXtreme gradient boosting (XGBoost) against the ensemble approach of the top two performed models. We hypothesized that the proposed ensemble ML approach enhances classification accuracy by integrating the advantages of top-performing models. Our proposed ensemble approach achieved a classification accuracy of 95.24%; this performance is significantly high compared to the stand-alone ML models. With the highest ROC value of 99.08%, our proposed ensemble approach shows its potential as a good learning model. Our work aims to discover the potential of the ensemble ML approach, which may be implemented with real-time data to assist physicians in improving diagnosis and improve clinical decision-making.