Exploring Relationships Between Maternal Characteristics, Continuous Glucose Monitoring Data, and Neonatal...

Area of Science:

• Reproductive Medicine
• Endocrinology
• Biostatistics

Background:

• Gestational diabetes mellitus (GDM) is a common pregnancy complication impacting maternal and neonatal health.
• Exploring maternal characteristics, wearable technology data (CGM), and neonatal outcomes can improve GDM management.
• Identifying risk factors for adverse neonatal outcomes (NH) in GDM pregnancies is crucial.

Purpose of the Study:

• To analyze interactions between maternal factors, CGM data, and neonatal outcomes in GDM pregnancies.
• To develop predictive models for neonatal complications using Bayesian networks.
• To identify key variables influencing neonatal outcomes in GDM.

Main Methods:

• Utilized Bayesian networks (BNs) for probabilistic modeling in the STEADY SUGAR trial (N=118 GDM patients).
• Analyzed maternal characteristics, CGM-derived features (90 days pre-delivery), and neonatal outcomes (NH).
• Quantified statistical relevance using odds ratios (ORs) and conditional probability tables.

Main Results:

• Direct associations found between NH and maternal hypertension (OR: 2.13), family history of diabetes (OR: 1.43), and elevated BMI (OR: 3.59).
• Cesarean delivery (OR: 2.05) and medication regimens indirectly influenced NH risk.
• Significant ethnic disparities observed, with higher hyperglycemia in Afro-American patients (OR: 2.91); CGM metrics linked to neonatal outcomes.

Conclusions:

• Bayesian networks effectively explored complex variable interactions in GDM pregnancies.
• The framework provides insights for clinical decision-making to mitigate maternal and neonatal risks.
• Future work will involve larger datasets to enhance predictive capabilities for GDM management.