ALT

This study aimed to develop a predictive model for gestational diabetes mellitus (GDM) using a combination of clinical characteristics and serum biomarkers. Data were collected from 378 pregnant women with full-term singleton vaginal deliveries, and predictors were identified through stepwise logistic regression. The final model included eight factors: serum ALT, TBA, TC, TG levels, and personal characteristics such as appetite changes due to GDM, husband-wife relationship, family history of diabetes, and parental relationships. The model demonstrated good fit in the Hosmer-Lemeshow test (P > 0.05) and achieved an area under the ROC curve (AUC) of 0.93, 0.72, and 0.68 for the training, internal validation, and external validation sets, respectively. These findings suggest that the model, combining routine obstetric tests and psychosocial factors, may serve as an effective tool for early GDM risk prediction.