A Machine Learning-Based Framework for Multi-Class Prediction of Hepatitis C Severity Using Ensemble Techniques

Hepatitis C, induced by the hepatitis C virus (HCV), represents a major public health concern due to its potential to lead to severe liver complications like fibrosis, cirrhosis, and liver cancer. Without a vaccine for chronic Hepatitis C, early diagnosis and prompt treatment are crucial. Traditional diagnostic methods are often time-consuming, costly, and prone to false negatives, especially in early infection stages. This study addresses these issues by introducing a machine learning-based multi-class classification framework to predict Hepatitis C severity. Using laboratory data from blood donors and patients, the study employed KNN imputation, the Adaptive Synthetic Sampling (ADASYN), and min-max normalization for data preparation. Ensemble learning methods, including voting, bagging, and boosting, were used for classification, with Bayesian optimization and K-Fold cross-validation for model validation. According to the findings Random Forest model achieved ۹۹% accuracy, highlighting 'Aspartic Amino-Transferase' (AST) and 'Bilirubin' (BIL) as key predictors in the prediction of hepatitis C severity. These methods enhance the reliability of Hepatitis C severity prediction and offering a robust tool for early diagnosis.

کلیدواژه ها:

Hepatitis C ، Machine Learning ، Multi-Class Classification ، Ensemble Techniques ، Adaptive Synthetic Sampling ، Healthcare System