Predicting gestational diabetes before conception for personalized interpregnancy weight management

Abstract

Abstract The growing recognition of the importance of interpregnancy care to reduce gestational diabetes mellitus (GDM) risk underscores the importance of effective preventive strategies. However, developing effective systems is still challenging. We aimed to bridge this gap by developing a weight management specific prediction model. This study retrospectively analyzed the data of women who underwent two childbirths across 15 medical facilities, including both primary and tertiary facilities. A derivation cohort was constructed using data from 2009 to 2019 (n = 1,640). Data between 2020 and 2024 was used to construct a separate temporal-validation cohort (n = 293). Using the data from another tertiary center between 2017 and 2023, the geographical-validation cohort was constructed (n = 339). A prediction model for GDM development in the second pregnancy was developed by applying logistic regression analysis using 5 key clinical information. GDM in the second pregnancy occurred in 9.5% (156 of 1,640, derivation), 16.7% (49 of 293, temporal-validation), and 7.7% (26 of 339, geographical-validation). The prediction model demonstrated consistent discrimination across cohorts, with c-statistics of 0.75, 0.80, and 0.79, respectively. Precision–recall analyses, accounting for the low prevalence of GDM, further confirmed performance well above the baseline (0.095 in the derivation cohort), with AUC-PRs of 0.43, 0.47, and 0.41 for the three cohorts. Calibration showed alignment with slopes of 1.04, 0.87, and 0.59 for each cohort. This simple and accurate model supports personalized weight management goals, offering a practical tool to reduce GDM risk in future pregnancies through inter-pregnancy weight management.