Integrating Machine Learning and In Silico Analysis for Enhanced Detection and Treatment of NAFLD-Linked Hepatocellular Carcinoma

Hepatocellular Carcinoma (HCC), which ultimately arises from Non-Alcoholic Fatty Liver Disease (NAFLD), is a poignant challenge for individuals suffering from various metabolic disorders, and the occurrences have increased over the last twenty years. Prompt recognition and accurate therapies can be vital for improving survival rates; however, traditional approaches do not provide the requirements needed to address the unique challenges associated with HCC advancement, which originate from Hepatitis B and C viral susceptibility, fatty liver disease, genetic changes, and immune disorders. Innovative approaches through machine learning (ML) and artificial intelligence (AI) could enhance HCC early intervention through timely diagnosis and research efforts. They have demonstrated significant effectiveness by employing techniques like histopathological analysis and radiomics to analyze medical images and tissue samples with outstanding accuracy for other prevalent cancers. For NAFLD-associated HCC, they can aid in differentiating its treatment from alcohol-related liver cancer, as AI models can identify genetic and molecular markers to develop targeted therapies, improving treatment efficacy and minimizing side effects of drugs, and ML can assist in discovering potential therapies by analyzing extensive datasets of molecular and biological interactions, significantly decreasing development times and costs. This study intends to explore the effectiveness of AI and ML methods in determining accuracy of HCC analysis, where for two datasets of NAFLD and HCC, the ML algorithms XGBoost and Random Forest gave better accuracy of 93% and 81.6% respectively. Also, through in silico analysis like molecular docking of possible drug ligands of HCC interacting with associated biomarker proteins and the toxicity evaluation of the drug ligands, it was found that the interaction of the tumor suppressor protein TP53 with the novel drug ligand Resmetirom gave best binding affinity out of other protein-ligand combinations with the score of -11.22 kcal/mol. While further analysis of this disease is necessary, computational techniques can offer a glimpse into targeting specific genes or proteins associated with HCC and can help in developing more personalized and effective therapies for NAFLD-related HCC in the future.