Gestational Diabetes Mellitus (GDM) is a condition diagnosed between the 24th and 28th week of pregnancy, excluding pre-existing diabetes or early pregnancy glucose abnormalities. According to the International Diabetes Federation, the global prevalence of GDM was 8.6% in 2021. In China, the prevalence has risen from 4% in 2010 to 21% in 2020, attributed to economic development, lifestyle changes, delayed childbearing age, and evolving reproductive beliefs.
GDM increases maternal and fetal risks, including preeclampsia, cesarean delivery, shoulder dystocia, macrosomia, and congenital malformations. It also raises long-term risks of type 2 diabetes, cardiovascular diseases, and chronic kidney disease in women. Effective management of GDM is critical for ensuring maternal and fetal health. Guidelines from the Chinese Medical Association, American Diabetes Association, and Queensland Health recommend lifestyle modifications (diet and exercise) as first-line treatment. However, 20-30% of women with GDM fail to achieve blood glucose control through lifestyle changes alone, necessitating pharmacological interventions such as insulin or oral hypoglycemic agents.
This study aims to develop a predictive model using machine learning algorithms based on individualized pre-pregnancy, pregnancy, and post-diagnosis data to identify women at risk of ineffective lifestyle management. The model will provide early warning signals, assisting healthcare providers in identifying those who require more intensive treatment and enabling timely adjustments to care plans. Key inputs for the model include personalized clinical data, biomarkers, lifestyle, and behavioral factors.
By leveraging machine learning, this study intends to build a more precise predictive tool to optimize GDM management and improve maternal and fetal outcomes.