Non-Alcoholic Fatty Liver Disease (NAFLD) is currently the most common cause of long-term liver disease worldwide, and more and more people are affected by it every year. As the disease progresses and turns into nonalcoholic steatohepatitis (NASH) the scarring and damage to the liver increases, which has been associated with a heightened risk of liver cancer, heart disease and overall mortality. Most patients do not show any symptoms in the early stages of the disease, which means that often diagnosis happens too late for effective treatment. In addition, accurate diagnosis today is only possible by taking a liver biopsy, which requires an invasive procedure. However, most patients suspected to have NAFLD/NASH are not confirmed. This means a large number of people unnecessarily undergo an invasive procedure and early diagnosis of non-alcoholic steatohepatitis (NASH) is often not possible. Accurate diagnosis of fibrosis risk is crucial for better treatment and management of the disease. In this project we aim to use state-of-the-art interpretable machine-learning technology in combination with one of the richest health datasets in the world to find new biomarker signatures that could help us develop non-invasive tests that could diagnose patients earlier and more accurately than is done today.
