Background/Objectives: Being overweight in childhood is a strong predictor of later obesity and related health disorders, underscoring the importance of the early identification of at-risk children. The aim of this study was to develop a prediction model for being overweight across childhood and adolescence using routine life-course health records available up to 18–23 months of age. Methods: We analyzed 1581 participants from the Tohoku Medical Megabank Birth and Three-Generation Cohort Study and constructed multivariable logistic regression models to predict being overweight at 36–47 months, 6 years, 11 years, and 14 years. Predictors included being overweight at 18–23 months, maternal characteristics, birth weight, and body mass index changes from birth to 18–23 months. Model performance was evaluated using the area under the curve (AUC), calibration, Brier scores, risk by score range, and stratified 10-fold cross-validation to assess the stability and robustness of predictive performance. Results: Being overweight at 18–23 months consistently showed strong associations with later overweight status. Model discrimination was moderate to high for early and middle childhood (AUC 0.873 at 36–47 months; 0.772 at 6 years) but modest for adolescence (AUC 0.720 and 0.692 at 11 and 14 years). Cross-validation demonstrated stable predictive performance across all age groups. Calibration and overall predictive accuracies were acceptable across all age groups. Conclusions: These results indicate that routine early life-course health records moderately predict the risk of being overweight, supporting their practical potential for early preventive interventions.
Shinoda et al. (2026) studied this question.