Predicting neurodevelopmental outcomes in neonatal hyperbilirubinemia: a multidimensional nomogram integrating biomarkers and neurobehavioral scores

Objective: The long-term neurodevelopmental trajectories of neonates with hyperbilirubinemia lacking overt acute symptoms remain largely unknown. This study aimed to evaluate these trajectories and develop a predictive model integrating multidimensional biomarkers with early neurobehavioral screenings to forecast neurodevelopmental impairment at one year of age. Methods: In this prospective observational study of 234 term neonates, clinical characteristics were recorded at admission, and laboratory results were obtained within the first 24 h. Early Neonatal Behavioral Neurological Assessment (NBNA) was performed, and neurodevelopment was assessed at 12 months of age using the Gesell Developmental Schedules. Key predictors were selected by integrating LASSO regression results with prior knowledge and clinical significance. A multivariable logistic regression model was subsequently constructed to develop a prognostic nomogram, internally validated using bootstrap resampling. Results: At the 12-month follow-up, 12.8% (30/234) of the infants exhibited neurodevelopmental impairment. Multivariable analysis identified elevated C-reactive protein (CRP, OR = 1.097, 95% CI: 1.026, 1.172), a higher bilirubin-to-albumin ratio (B/A, OR = 2.587, 95% CI: 1.497, 4.471), and increased neuron-specific enolase (NSE, OR = 1.285, 95% CI: 1.092, 1.513) as independent risk factors for adverse outcomes. Higher NBNA scores (OR = 0.749, 95% CI: 0.608, 0.922) served as an independent protective factor. The model yielded an apparent area under the curve (AUC) of 0.813 and an optimism-corrected AUC of 0.797 (95% CI: 0.696-0.902), demonstrating stable discriminative power. Calibration plots showed consistency between predicted and observed probabilities, with a Brier score of 0.09. Decision curve analysis indicated clinical net benefit across a range of threshold probabilities. Conclusion: Neonates with hyperbilirubinemia may experience subclinical neurodevelopmental impairment. Our model facilitates risk stratification, supporting targeted long-term follow-up and developmental monitoring for high-risk infants.