A Study on Improving Liver Cancer Staging Using Machine Learning Feature Selection

Hepatocellular carcinoma (HCC) is one of the deadliest tumors in the world, and treatment of HCC requires accurate staging and risk assessment of patients in order to develop individualized treatment plans and predict their prognosis. Therefore, clinical staging is closely related to patient outcomes and prognosis. However, traditional staging diagnosis relies on a single clinical index and is limited to liver function assessment. To address these issues, we used the least absolute shrinkage and selection operator (Lasso) for clinical characterization, applied COX survival analysis to generate a unique risk score for each patient, and finally used clustering to generate the appropriate staging system. The results show that our proposed staging system provides better results than the traditional staging system (TNM) for survival prediction, contains more features, and is able to assess the prognosis of patients in a more comprehensive way, thus providing more predictive value for personalized treatment and management.