Machine Learning-Based Liver Diseases Prediction: A Comparative Analysis

Liver disease contributes to a significant number of deaths worldwide, making it a public health concern that requires attention in terms of prevention, early detection, and effective management. The early detection of liver diseases is a critical area in medical diagnostics where early intervention can significantly alter patient outcomes. This research focuses on using machine learning (ML) approaches to predict a liver disease through accessible blood test data, aiding in early diagnosis and treatment. This study employed four ML algorithms: Random Forest, Support Vector Machine (SVM), Logistic Regression and XGBoost applying them to a dataset from the UCI repository and real dataset from Andalusia Group Hospitals in Egypt. It compared their performance using metrics like confusion matrix, precision, recall, F1 score, accuracy, ROC_AUC score. Following the data cleaning and preprocessing steps, by using the dataset splitting 90% training and 10% testing, the Random Forest and XGBoost showed the highest accuracy and AUC among the evaluated models and have the potential to serve as powerful instruments for healthcare practitioners. Utilizing common and cost-effective blood test data, these tools can proficiently aid in the prediction of liver disease.