Hepatitis C is an infection of the liver caused by the hepatitis C virus (HCV) [1]. Although acute hepatitis C infection may result in spontaneous clearance in 15 % of cases, it becomes chronic in 85 % of cases. Chronic hepatitis C (CHC) infection may result in serious complications such as cirrhosis and hepatocellular carcinoma in the long term [2]. Furthermore, hepatitis is a significantly underdiagnosed infection with the vast majority of the estimated 71 million people living with HCV believed to remain undiagnosed and untreated. Thus, the World Health Organization (WHO) recognizes viral hepatitis as a major public health threat and has implemented hepatitis elimination programs [3]. The high (>90 %) cure rates achieved with direct-acting antiviral (DAA) agents in the treatment of active HCV infection, have made HCV elimination a more attainable target [4], [5], [6]. However, as no vaccine that has been developed as yet and the risk of re-infection continues even in patients who have been cured with antiviral treatment, the only way to eliminate HCV is to screen large masses for the presence of viremia (active HCV infection) as soon as possible and to treat viremic cases promptly to minimize the risk of transmission [5], [7].

The diagnosis of current (active) HCV infection consists of two stages. In the first stage, serological hepatitis C antibodies (anti-HCV) assays (using ELISA or CIA methods) are usually used as a routine screening test to identify specific antibodies to HCV. If the result is anti-HCV positive (S/Co ≥ 1), it should be followed by a molecular assay to detect HCV RNA in the second stage. At this stage, the presence of HCV viremia (active HCV infection) can be confirmed with a positive HCV RNA assay result, or rejected with a negative result [8], [9], [10]. However, the HCV RNA assay is more expensive and requires time and trained personnel. Moreover, in areas of low incidence (<10 %), high rates (35–60 %) of false-positive anti-HCV results, have caused a large number of unnecessary investigations to be conducted, wasting both time and already limited economic resources [11], [12]. Therefore, the HCV RNA assay cannot be used for routine surveillance of the presence of viremia as it is not suitable for use in a global elimination program. Taken together, these findings heighten the need for more effective alternative approaches and methods to diagnose viremia earlier.

Machine learning (ML) techniques are methods that have become prominent in the field of healthcare as they have shown higher performance in the correct prediction of the dependent variable and in extracting meaning from the data, with the inclusion of many more independent variables compared to the hypothesis-based classical statistical methods [13], [14]. From this perspective, the development of an ML algorithm to provide an early and accurate prediction of the presence of viremia, without waiting for HCV RNA result, in anti-HCV seropositive cases, would enable interventions to reduce or prevent disease transmission to be applied faster in larger populations.

To the best of our knowledge, no such ML model has been developed on this subject. Hence, the aim of the current study was to design a new ML model to predict the presence of viremia in anti-HCV seropositive cases.