Scholarlink

<p>Hepatitis C is a liver disease caused by the Hepatitis C virus (HCV) that can progress from acute to chronic stages. Its early detection is necessary so that timely treatment can be provided and progression from acute to chronic stage can be reduced. This paper presents an innovative explainable AI-driven framework Hepa-HISS (Hepatitis Hybrid Intelligent Stage Screening), for Hepatitis C detection by integrating hybrid feature selection, polynomial interaction feature generation, and boundary-aware synthetic minority oversampling technique (BSMOTE) with gradient boosting. A hybrid feature selection technique ANORFE is introduced by integrating analysis of variance (ANOVA) and recursive feature elimination (RFE) methods. Non-linear interaction between features is extracted using polynomial interaction features. Class imbalance is addressed using BSMOTE. Multiple classifiers including Random Forest (RF), Gradient Boosting (GB), Support Vector Machine (SVM), k-Nearest Neighbor (KNN), Extreme Gradient Boosting (XGBoost), and Light Gradient Boosting Machine (LGBM) are evaluated within the proposed framework. The presented Hepa-HISS framework performs best with Random Forest, achieving an accuracy of 96.58%, by using nested cross-validation. Several evaluation metrics, such as precision, recall, F1-score, Matthews Correlation Coefficient (MCC), and Cohen's Kappa, were also examined as dataset is highly imbalanced. The results show that the presented framework performs consistently across both majority and minority classes, indicating its effectiveness for Hepatitis C stage prediction.SHAP and LIME analyses are also applied to add explainability to the model outputs.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An explainable machine learning framework integrating hybrid feature selection and interaction modeling for hepatitis C stage prediction

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月19日

查看全文
  • Rohit Lamba,
  • Pooja Rani,
  • Ravi Kumar Sachdeva,
  • Priyanka Bathla,
  • Anurag Jain,
  • Bhupesh Kumar Dewangan

摘要

Hepatitis C is a liver disease caused by the Hepatitis C virus (HCV) that can progress from acute to chronic stages. Its early detection is necessary so that timely treatment can be provided and progression from acute to chronic stage can be reduced. This paper presents an innovative explainable AI-driven framework Hepa-HISS (Hepatitis Hybrid Intelligent Stage Screening), for Hepatitis C detection by integrating hybrid feature selection, polynomial interaction feature generation, and boundary-aware synthetic minority oversampling technique (BSMOTE) with gradient boosting. A hybrid feature selection technique ANORFE is introduced by integrating analysis of variance (ANOVA) and recursive feature elimination (RFE) methods. Non-linear interaction between features is extracted using polynomial interaction features. Class imbalance is addressed using BSMOTE. Multiple classifiers including Random Forest (RF), Gradient Boosting (GB), Support Vector Machine (SVM), k-Nearest Neighbor (KNN), Extreme Gradient Boosting (XGBoost), and Light Gradient Boosting Machine (LGBM) are evaluated within the proposed framework. The presented Hepa-HISS framework performs best with Random Forest, achieving an accuracy of 96.58%, by using nested cross-validation. Several evaluation metrics, such as precision, recall, F1-score, Matthews Correlation Coefficient (MCC), and Cohen's Kappa, were also examined as dataset is highly imbalanced. The results show that the presented framework performs consistently across both majority and minority classes, indicating its effectiveness for Hepatitis C stage prediction.SHAP and LIME analyses are also applied to add explainability to the model outputs.