A New Machine Learning-Based Complementary Approach for Screening of NAFLD (Hepatic Steatosis)

Non-Alcoholic Fatty Liver Disease (NAFLD) is the major reason for liver disease globally. Early warning of liver disease at the beginning of a progressive disease spectrum is critical for reduced mortality and increased longevity. Current clinical practices focus on disease management but can be improved in terms of screening & early detection. This paper focuses on machine learning-based intelligent model development using liver functionality and physiological parameters for Hepatic Steatosis (Non-alcoholic Fatty Liver) screening. Gender-specific models were developed separately. Customized data processing techniques were incorporated. Publicly available, population data (NHANES-III) was used. The maximum sensitivity provided by the models were approximately 72% and 71% for male and female, respectively. Maximum specificities obtained by the models were 74% and 75% for male and female, respectively. Performance comparison of different models has been discussed.