Construction and Evaluation of a Prediction Model for Neonatal Hyperbilirubinemia.

Objective: To investigate the factors that contribute to neonatal hyperbilirubinemia (NHB), develop a prediction model and predictive factors, which provide reference base for the early detection of NHB. Methods: A retrospective study was done to collect clinical data of 683 neonates and their mothers which delivered in the hospital's obstetrics department between January 2019 and January 2023, of these neonates 216 were hyperbilirubinemic and 467 were not. Clinical data and laboratory results of newborns and their mothers were collected and analysed. Multifactorial logistic regression analysis was utilised to develop an early prediction model and identify determinants of newborn hyperbilirubinemia. Results: The logistic regression model revealed that Delayed meconium (OR=4.024, P=0.002), maternal age (OR=1.106, P<0.01), maternal-infant blood group incompatibility (OR=4.457, P=0.001), Gestational Diabetes Mellitus (GDM) (OR=5.356, P=0.03), Pre-pregnancy co-morbidity (OR=2.810, <0.01), and Weight Growth Rate (WGR) during pregnancy (OR=28.367, P=0.048) were independent risk factors for NHB; 25-(OH)D3 (OR=0.880, P=0.002). The Conjoint predictor ROC curve is below 0.851 (P<0.01, 95% CI:0.821-0.882). The highest Youden's index was 0.55, with a sensitivity of 0.81 and a specificity of 0.76, indicating that the predictor had a decent predictive effect. Conclusion: This study identified the risk factors and protective factors associated with NHB. Additionally, a joint prediction model was developed to more accurately predict the risk of hyperbilirubinemia which can serve as a foundation for the clinical identification of pertinent susceptibility factors and the development of intervention measures.