Hybrid Model for Prediction of Treatment Response in Beta-thalassemia Patients with Hepatitis C Infection

Abstract

Patients with beta-thalassaemia major (β-TM) who get regular blood transfusions are at risk of iron overload and hepatitis C virus (HCV) infection. These double injuries together can lead to chronic liver damage. Treatment with pegylated interferon combined ribavirin (Peg-IFN/RBV) is associated with side effects that compromise the patients' quality of life. The efficacy of two anti-viral regimens (Peg-IFN/RBV) and Peg-IFN monotherapy were assessed using a machine learning model to identify patients who could achieve sustained virologic response (SVR) with HCV eradication. This paper is a follow-up study of our previous published paper that used a different method to address the same research question. A hybrid Neuro-SVM model was developed to improve the accuracy of classification that shows 98.83% in group 1 and 99.75 in group 2 and conveyed as a graphical user interface that can help the clinical support decision in the prediction of optimal treatment response. The model was compared to artificial neural network (ANN), support vector machine (SVM) and naïve Bayesian (NB). Using the hybrid model, it would be useful if we distinguish in advance those patients who may benefit from the approved direct anti-viral agents (DAAs) therapy from those who would not. Keywords Beta-thalassemia major • Synthetic minority oversampling • Machine learning • Artificial neural networks All authors assisted with manuscript preparation and revisions. AH and AA designed the study; SK provides the dataset with laboratory follow-up of enrolled patients. AH analyzed the data and wrote the manuscript. All authors read and approved the final version of the manuscript.