The Construction and Performance Evaluation of a Risk Prediction Model for Nonalcoholic Steatohepatitis Based on Serological Markers

Background and Aims To develop a noninvasive clinical diagnostic model based on serological markers for nonalcoholic steatohepatitis (NASH) and to verify its predictive efficacy. Methods A total of 82 biopsy‐proven patients with nonalcoholic fatty liver disease (NAFLD) were included in the study. Patients were classified into nonalcoholic fatty liver (NAFL) and NASH groups based on the results of liver biopsies. The study utilized the LASSO regression model for variable selection, followed by logistic regression analysis to create a prediction model. A nomogram was then developed to illustrate this model. To validate the model, bootstrapping was applied for internal validation, and the accuracy, consistency, and clinical utility of the prediction model were evaluated. Results The NASH group had significantly higher levels of red blood cell count, lactate dehydrogenase (LDH), aspartate aminotransferase (AST), and alanine aminotransferase (ALT), while levels of high‐density lipoprotein (HDL) cholesterol were significantly lower in the NASH group ( p < 0.05). Logistic regression analysis indicated that AST and ceruloplasmin were independent risk factors associated with NASH. A nomogram based on serological markers, including ceruloplasmin, HDL, AST, red blood cell count, thyroid‐stimulating hormone (TSH), and total bile acid (TBA), was established to predict NASH with excellent discrimination (AUROC 0.813). Conclusions AST and ceruloplasmin are independent risk factors associated with NASH. The CHART2 prediction model based on serological markers demonstrates good accuracy, consistency, and clinical utility. The model could serve as a noninvasive approach to identifying patients with NASH, which might reduce the need for liver biopsy.