ENHANCING THE PERFORMANCE OF MACHINE LEARNING MODELS FOR THE DIAGNOSIS OF LIVER DISEASE

Abstract

Liver diseases represent a major global health burden, contributing to millions of deaths annually due to delayed diagnosis and inadequate clinical screening mechanisms. Early and reliable identification of liver disorders is therefore critical to improve patient outcomes and reduce healthcare costs. This study proposes an optimized machine learning (ML)-based diagnostic framework to enhance predictive performance using systematic preprocessing, dataset balancing, and hyperparameter tuning. The Indian Liver Patient Dataset (ILPD) from the UCI Machine Learning Repository was employed to evaluate several ML models, including Random Forest (RF), Support Vector Machine (SVM), k-Nearest Neighbor (kNN), and Gradient Boosting (GB). Rigorous data preprocessing involved duplicate removal, missing value imputation using Multivariate Imputation by Chained Equations (MICE), Z-score standardization, and outlier elimination. Synthetic Minority Oversampling Technique (SMOTE) was applied to address class imbalance, while GridSearchCV and RandomizedSearchCV were used for hyperparameter optimization. The optimized Random Forest model achieved the highest accuracy of 84.52%, outperforming other classifiers in precision (90.33%), recall (81.81%), and F1-score (85.86%), with a statistically significant p-value of 1.21×10⁻¹⁶. The findings underscore the effectiveness of model optimization and balanced data handling in improving diagnostic accuracy for liver disease. The proposed approach provides a robust foundation for intelligent decision-support systems in clinical environments and paves the way for further integration of data-driven methodologies in hepatology.