Abstract
To address the need for a simple model to predict ≥ F2 fibrosis in metabolic dysfunction-associated steatotic liver disease (MASLD) patients, a study utilized data from 791 biopsy-proven MASLD patients from the NASH Clinical Research Network and Jinan University First Affiliated Hospital. The data were divided into training and internal testing sets through randomized stratified sampling. A multivariable logistic regression model using key categorical variables was developed to identify ≥ F2 fibrosis. External validation was performed using data from the FLINT trial and multiple centers in China. The DA-GAG score, incorporating diabetes, age, GGT, aspartate aminotransferase/ platelet ratio, and globulin/ total protein ratio, demonstrated superior performance in distinguishing ≥ F2 fibrosis with an area under the receiver operating characteristic curve of 0.79 in training and over 0.80 in testing datasets. The DA-GAG score efficiently identifies MASLD patients with ≥ F2 fibrosis, significantly reducing the medical burden.