Liver hepatocellular carcinoma (LIHC) is a prevalent and highly aggressive form of liver cancer, characterized by increasing rates of incidence and mortality globally. Although numerous treatment options currently exist, they frequently result in insufficient clinical outcomes for those diagnosed with LIHC. This highlights the urgent need to identify new biomarkers that can enhance prognostic evaluations and support the development of more effective therapeutic strategies for LIHC. Through the use of the SwissTargetPrediction tool, we precisely identified molecular targets related to Sorafenib. Furthermore, analysis of RNA sequencing data from the TCGA-LIHC cohort uncovered 24 genes associated with different patient prognoses following Sorafenib therapy. Employing a clustering-based analytical approach, we assessed the connections between gene expression profiles, clinical outcomes, immune cell infiltration levels, and tumor stage progression. A prognostic framework was constructed by applying various machine learning techniques and subsequently validated across several independent datasets. Utilizing the XgBoost algorithm, MAPK12 emerged as a key regulatory gene influencing the prognosis of individuals with LIHC. The results of in vitro experiments demonstrated that knockdown of MAPK12 reduced the proliferation, metastasis, and tumor stemness-related sphere-forming ability of LIHC cells. These results underscore the promise of MAPK12 as a potential prognostic biomarker for LIHC and offer valuable insights for crafting personalized treatment approaches.