PROGNOSTIC MODELING FOR HEPATIC DISORDERS: A PARADIGM OF EQUILIBRATED AND GENERALIZED MACHINE LEARNING METHODOLOGIES

Abstract

In human physiology, the liver is a vital organ responsible for performing critical functions such as bile production, bilirubin excretion, metabolism of proteins and carbohydrates, enzyme activation, glycogen storage, and plasma protein synthesis. However, it is highly susceptible to damage due to alcohol consumption, certain medications, and poor dietary habits. Traditional diagnostic methods for liver disorders, including blood tests and imaging, are time-consuming and costly, often delaying crucial treatment. This study introduces a machine learning-based prognostic framework to enhance the speed and accuracy of liver disease diagnosis. The proposed approach integrates advanced algorithms, including Random Forest, Gradient Boosting, XGBoost, and LightGBM, combined with an ensemble voting method to leverage their complementary strengths. Preprocessing techniques such as Principal Component Analysis (PCA) for dimensionality reduction and Synthetic Minority Oversampling Technique (SMOTE) to address class imbalance were employed to refine the dataset. Evaluation metrics like precision, recall, F1-score, accuracy, and ROC-AUC revealed the ensemble model’s superior performance, achieving the highest accuracy of 98% and a ROC-AUC of 0.9963, significantly outperforming individual models. This study offers a scalable and cost-effective solution that reduces diagnostic time and improves predictive reliability. The framework provides significant advantages for medical applications, serving as a decision support tool to aid healthcare professionals in timely and accurate liver disorder diagnosis, particularly in resource-limited settings.