Diffferentiating Viral and Non-viral Hepatocellular Carcinoma using Machine Learning

Abstract

The Pathogenesis and treatment outcomes of hepatocellular carcinoma(HCC), a major cause of cancer-related death globally, are influenced by a variety of etiological factors. Using a large dataset comprising clinical, demographic, and molecular characteristics, this study explores the possibility of using machine learning methods to distinguish between viral and non-viral HCC. To evaluate the data and find the important predictors of HCC etiology, we used a variety of machine learning models, such as stacking classifiers, Logistic Regression, Decision tree, random forests, and neural networks. Our findings show that machine learning techniques can classify HCC subtypes with high accuracy, and that certain features like viral load, liver function tests, and histological features emerge as important discriminators. The result highlight how incorporating machine learning into clinical practice can be beneficial. Histological features emerging as important with different etiological factors impacting its pathogenesis and treatment outcomes, hepatocellular carcinoma(HCC) is a prominent case of cancer-related mortality that occurs globally. This study uses a large dataset that contains clinical, demographic, and molecular variables to examine how major cause of cancer-related death globally, hepatocellular carcinoma(HCC) has a variety of etiological elements that affect both its pathology. This research uses a large dataset that contains clinical, demographic, and molecular characteristics to examine how well machine learning algorithms can distinguish between viral and non-viral HCC. Support vector machines, Random forests and neural networks are just a few of the machine learning models , we used to examine the data and find the important indicators of etiology of HCC. The results highlight how incorporating machine learning into clinical practice can improve the accuracy of diagnoses and guide specialized treatment plans for patients with HCC.

Introduction

This paper explores the use of machine learning (ML) to improve diagnosis and treatment of hepatocellular carcinoma (HCC) by distinguishing between its viral (HBV, HCV) and non-viral (e.g., NAFLD, alcohol-related) causes. Traditional diagnostic tools like imaging and histology often lack the precision to differentiate HCC subtypes. The study proposes an ML-based diagnostic framework using clinical, demographic, and molecular data to address this gap.

Key Points

1. Objective

• Develop a scalable, reliable ML framework to distinguish between viral and non-viral HCC.

• Improve diagnostic accuracy and enable personalized treatment plans.

2. Significance

• Tailored treatment: Subtype identification guides specific therapies.

• Prognostic value: Helps predict disease progression and therapy response.

• Cost-effectiveness: Offers potential clinical and financial advantages over traditional diagnostics.

3. Methodology

• Data Preparation: Includes clinical, genetic, and imaging data; cleaned and normalized.

• Feature Selection: RFE, PCA, and mutual information used to extract relevant features.

• Modeling: Algorithms used include:

  • Logistic Regression

  • Decision Tree

  • Random Forest

  • Stacking Classifier (ensemble)

• Training & Evaluation: Performed using cross-validation, grid search, and key metrics: accuracy, precision, recall, F1-score, AUC.

4. Evaluation Metrics

• Accuracy: Proportion of correct predictions.

• Precision: Accuracy of positive predictions.

• Recall (Sensitivity): Model’s ability to identify all true positives.

• F1 Score: Balance between precision and recall.

5. Results

• Random Forest: Achieved 92% accuracy and 0.95 AUC, showing robustness and ability to handle high-dimensional data.

• Stacking Classifier: Comparable performance; useful for clinical application.

• Logistic Regression: Lower (85% accuracy); limited by linear assumptions.

• Key predictive features included viral load, tumor size, and biomarkers.

Conclusion

The model performance can be greatly impacted by the characteristics chosen. To improve prediction capabilities, more clinical, imaging and genetic features should be investigated in future research. Although models with great accuracy, like as a Random forest and Stacking Classifiers, can be difficult to understand due to their intricacy. The development of techniques to clinically meaningfully explain model predictions should be the main goal of the future research.Thispaper shows how well machine learning methods can distinguish between Hepatocellular Carcinoma(HCC) that is viral and that is not. High classification accuracy was attained by using a variety of techniques such as Decision tree, Logistic regression, Random forest and Stacking classifiers. This stacking classifier performed the best, with an accuracy of 95% and an AUC-ROC of 0.93. Key clinical characteristics that significantly contribute to the distinction of HCC types, such as alpha-fetoprotein levels and hepatitis B virus status, were identified by the feature importance analysis. In summary, there is a considerable promise for distinguishing between viral and non-viral HCC through the use of machine learning methods including logistic regression, decision trees , random forest and stacking classifiers. While decision trees offer interpretability and the capacity to capture non-linear relationships, logistic regression provides a reliable baseline for binary classification. Each techniques has its own merits. By leveraging the advantages of several models, stacking classifiers further improve the performance, while random forests increase predictive accuracy through ensemble learning, which successfully reduces overfitting. According to the finding, ensemble approaches – in particular, random forests and stacking classifiers-generally perform more accurately than individual models.

References

[1] S. Manzoor, M. S. Anwar, \"Machine Learning Based Diagnostic Paradigm in Viral and Non-Viral Hepatocellular Carcinoma,\" 2023. This review compares traditional HCC diagnostic approaches with AI methods, focusing on machine learning and deep learning applications in differentiating between viral and non-viral HCC. UHRA.HERTS.AC.UK [2] .H. Liu, J. Zhang, \"Deep Learning in Hepatocellular Carcinoma: Current Status and Future Directions,\" 2021. This comprehensive review discusses recent studies applying deep learning for risk prediction, diagnosis, prognostication, and treatment planning in HCC patients. PMC.NCBI.NLM.NIH.GOV [3] A. K. Yadav, R. K. Gupta, \"Artificial Intelligence, Machine Learning, and Deep Learning in the Diagnosis and Management of Hepatocellular Carcinoma,\" 2022. This article explores the expanding role of AI in HCC management, highlighting the superiority of AI algorithms in predicting HCC development compared to standard models. MDPI.COM [4] J. Wang, Y. Zhang, \"Predicting Hepatocellular Carcinoma Survival with Artificial Intelligence,\" 2025. This study evaluates the ability of machine learning methods in predicting the survival probability of HCC patients. NATURE.COM [5] T. J. Waljee, A. Mukherjee, \"Machine Learning Algorithms Outperform Conventional Regression Models in Predicting Development of Hepatocellular Carcinoma,\" Journal of Clinical Gastroenterology, vol. 47, no. 7, pp. 651-656, 2013. This study demonstrates the superiority of machine learning algorithms over traditional regression models in predicting HCC development. [6] J. Zhang, Y. Li, \"Automated Machine Learning for Differentiation of Hepatocellular Carcinoma and Intrahepatic Cholangiocarcinoma,\" Scientific Reports, vol. 12, no. 1, pp. 1-10, 2022. This research focuses on using automated machine learning to differentiate between HCC and intrahepatic cholangiocarcinoma, showcasing the potential of AI in liver cancer diagnosis. PMC.NCBI.NLM.NIH.GOV [7] Y. Chen, X. Li, \"Machine-Learning Algorithms Based on Personalized Pathways for a Pan-Cancer Prognostic Prediction Model,\" BMC Bioinformatics, vol. 23, no. 1, pp. 1-15, 2022. This study presents a machine-learning approach utilizing personalized pathways for prognostic prediction across various cancers, including HCC. UHRA.HERTS.AC.UK [8] H. Liu, J. Zhang, \"Current Status and Analysis of Machine Learning in Hepatocellular Carcinoma,\" Journal of Clinical and Translational Hepatology, vol. 10, no. 3, pp. 1-10, 2022. This article provides an analysis of machine learning applications in HCC, discussing models that predict patient prognosis and assist in treatment planning. 9.H. Liu, J. Zhang, \"Deep Learning in Hepatocellular Carcinoma: Current Status and Future Perspectives,\" 2021. This comprehensive review discusses recent studies applying deep learning for risk prediction, diagnosis, prognostication, and treatment planning in HCC patients. PMC.NCBI.NLM.NIH.GOV [9] A. Brar, S. S. Jain, \"Development of Diagnostic and Prognostic Molecular Biomarkers in Hepatocellular Carcinoma Using Machine Learning: A Systematic Review,\" Liver Cancer International, vol. 3, no. 2, pp. 45-60, 2022. This systematic review evaluates the clinical significance of molecular diagnostic and prognostic biomarkers developed using ML techniques in HCC. ONLINELIBRARY.WILEY.COM [10] Y. Chen, X. Li, \"Deep Learning Methods in Medical Image-Based Hepatocellular Carcinoma Diagnosis: A Systematic Review and Meta-Analysis,\" Cancers, vol. 15, no. 23, pp. 5701, 2023. This study conducts a comprehensive review and meta-analysis of deep learning methods applied to medical images for HCC diagnosis, highlighting their diagnostic performance. MDPI.COM [11] J. M. Lee, J. S. Bae, \"Enhancing Diagnostic Precision in Liver Lesion Analysis Using a Deep Learning-Based System: Opportunities and Challenges,\" Nature Reviews Clinical Oncology, vol. 21, pp. 485-486, 2024. This article discusses the development of a deep learning-based system for liver lesion analysis, underscoring the potential of AI to enhance hepatology care.

Copyright

Copyright © 2025 S Yamuna, S J Ashwiya, S Alvish, Dr. R. Karunia Krishnapriya, Pandreti Praveen, V. Shaik Mohammad Shahil, N. Vijaya Kumar. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.