A machine learning-based algorithm to improve surveillance of hepatocellular carcinoma in patients with liver cirrhosis” CORIS 2024-2025 - CUP: I93C25000380007

Chronic liver diseases and hepatocellular carcinoma (HCC) are a major public-health burden; HCC is often diagnosed late owing to the limitations of ultrasound surveillance and alpha-fetoprotein. The project develops and validates a machine-learning algorithm (SuperLearner, ensemble stacking) that, by integrating simple demographic, clinical and laboratory variables already available during follow-up, distinguishes cirrhotic patients with HCC from those with cirrhosis alone. The algorithm will be trained on a retrospective cohort (Padua), externally validated (Verona) and prospectively validated in three hepatology centres. The final aim is a free clinical decision-support application for risk stratification, promoting early diagnosis and improving patient prognosis.