Hepatitis C Prediction Using Feature Selection by Machine Learning Technique

1. Introduction

One of the main viruses that cause liver disease is the hepatitis C virus (HCV), which belongs to the Flaviviridae family. Approximately 175 million individuals globally, or 3% of the global population, are infected with HCV. Although 90% of injectable drug users are most at risk, parental transmission is the primary mode of HCV transmission. Conventional interferon and ribavirin, which have 38–43% sustained virological response rates, are still the gold standard for treating chronic HCV (Munir et al., 2010). Approximately 58 million people worldwide suffer from long-term hepatitis C virus disease, and 1.5 million new cases are reported each year. Nearly 3.2 million children and adolescents suffer from a chronic case of hepatitis C (WHO, 2023). It is asymptomatic at first, but when the infection worsens, it can cause chronic illnesses such as hepatocellular cancer and liver cirrhosis. To diagnose this illness, a few different non-invasive serum biochemical indicators are employed (Nandipati et al., 2020).

To determine the disease's stage, a variety of harmless blood biochemical indicators and patient medical information have been employed. Machine learning techniques have shown to be a helpful alternative for determining the phase of this chronic liver disease, avoiding the drawbacks of a biopsy (Butt et al., 2021). To stop the spread of disease and identify affected areas early on, medical research relies heavily on the forecasting and categorization of diseases. Machine learning (ML) techniques are frequently employed to accurately forecast and categorise diseases, serving as a useful tool for medical professionals (Mamdouh Farghaly et al., 2023). Clinical data contains complex and non-linear correlations that machine learning (ML) techniques are especially good at acquiring and analysing.

Through the identification of HCV-positive individuals, machine learning algorithms, including classification approaches, can be employed to create a model for HCV diagnosis. However, unsuitable attribute set features can degrade the classifier's effectiveness (John et al., 1994). When there are more important and non-redundant attributes in the data, various learning algorithms perform better and produce more accurate results. An effective feature selection strategy is required to extract intriguing aspects relevant to the condition, as clinical datasets contain a huge number of duplicated and irrelevant information (Jain & Singh, 2018).

Hepatocellular carcinoma (HCC) infection is still an important contributor to liver cirrhosis, liver transplants, and a global health problem today. But because of decades of incredible progress, HCV is now the first chronic viral infection that can be cured (Manns & Maasoumy, 2022) as shown in Table 1.