Inexpensive method for the quantitative estimation of hepatitis C virus RNA in blood plasma for low-resource settings using ML-based image intensity analysis of RT-LAMP products

Hepatitis C virus (HCV) infection is a severe public health problem affecting nearly 3% of the world population. Of those affected, approximately 80% develop chronic infections. Initiating treatment through HCV RNA testing remains challenging, especially in resource-limited settings where access to molecular diagnostics is restricted. In this study, a novel and inexpensive HCV molecular diagnostic approach based on on-chip reverse transcriptase loop-mediated isothermal amplification (RT-LAMP) integrated with image intensity measurement and machine learning-based prediction (RT-LAMP-IM-MLP) is developed. This method uses a low-cost microfluidic chip to carry out RT-LAMP (total cost of each test: <$4) and enables rapid, user-friendly, accurate, and quantitative detection of HCV RNA in blood plasma. Amplified products are visualized under fluorescence excitation and the captured image is processed using the OpenCV package in Python, followed by training and prediction through a modified random forest algorithm. When tested on plasma samples positive for HCV, hepatitis A virus (HAV), or from healthy individuals, the RT-LAMP-IM-MLP method demonstrates 97.1% sensitivity, 96.9% specificity, and 97% overall accuracy. Compared to the reference method, the real-time PCR-based COBAS® TaqMan® HCV assay (Roche Diagnostics, USA), our assay can detect HCV RNA concentrations as low as 10 IU mL^-1 (60 fg μL^-1). Further, only 0.5 μL of dye is used for fluorescence labeling compared to larger quantities required in colorimetric assays. Therefore, the proposed sensitive and specific detection scheme may serve as an inexpensive and reliable point-of-care (POC) test for detecting HCV RNA in clinical samples.