Comparative investigation of bagging enhanced machine learning for early detection of HCV infections using class imbalance technique with feature selection

• Published in: PloS one Vol. 20; no. 6; p. e0326488
• Main Authors: Tusher, Ekramul Haque, Ismail, Mohd Arfian, Akib, Abdullah, Gabralla, Lubna A., Ibrahim, Ashraf Osman, Som, Hafizan Mat, Remli, Muhammad Akmal
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 26.06.2025; Public Library of Science (PLoS)
• Subjects: Accuracy; Adult; Algorithms; Area Under Curve; Bagging; Bayes Theorem; Biology and life sciences; Care and treatment; Classification; Computer and Information Sciences; Control; Data mining; Data science; Datasets; Decision making; Decision Trees; Diagnosis; Disease; Early Diagnosis; Engineering and Technology; Ensemble learning; Feature selection; Female; Hepacivirus; Hepatitis C; Hepatitis C - diagnosis; Hepatitis C virus; Humans; Identification and classification; Infections; Learning algorithms; Liver cancer; Liver cirrhosis; Machine Learning; Male; Mean; Medical diagnosis; Medical research; Medicine and health sciences; Middle Aged; Oversampling; Physical Sciences; Recall; Research and Analysis Methods; Risk factors; Standard deviation; Support Vector Machine; Support vector machines; Variance analysis; Malaysia
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0326488

Around 1.5 million new cases of Hepatitis C Virus (HCV) are diagnosed globally each year (World Health Organization, 2023). Consequently, there is a pressing need for early diagnostic methods for HCV. This study investigates the prognostic accuracy of several ensemble machine learning (ML) models for diagnosing HCV infection. The study utilizes a dataset comprising demographic information of 615 individuals suspected of having HCV infection. Additionally, the research employs oversampling and undersampling techniques to address class imbalances in the dataset and conducts feature reduction using the F-test in one-way analysis of variance. Ensemble ML methods, including Support Vector Machine (SVM), k-Nearest Neighbors (k-NN), Logistic Regression (LR), Random Forest (RF), Naïve Bayes (NB), and Decision Tree (DT), are used to predict HCV infection. The performance of these ensemble methods is evaluated using metrics such as accuracy, recall, precision, F1 score, G-mean, balanced accuracy, cross-validation (CV), area under the curve (AUC), standard deviation, and error rate. Compared with previous studies, the Bagging k-NN model demonstrated superior performance under oversampling conditions, achieving 98.37% accuracy, 98.23% CV score, 97.67% precision, 97.93% recall, 98.18% selectivity, 97.79% F1 score, 98.06% balanced accuracy, 98.05% G-mean, a 1.63% error rate, 0.98 AUC, and a standard deviation of 0.192. This study highlights the potential of ensemble ML approaches in improving the diagnosis of HCV. The findings provide a foundation for developing accurate predictive methods for HCV diagnosis.