A novel nutritional tool to identify infants at risk of stunting

This study aimed to develop and validate a novel scoring system and prediction algorithm integrating growth, nutritional, and biochemical indicators for the identification of infant stunting, defined as length-for-age Z-score < -2 standard deviations according to the World Health Organization (WHO) Child Growth Standards.

A retrospective cohort of 380 infants (aged 0-12 months) undergoing routine health examinations was enrolled. Participants were randomly allocated to a training set (n = 266) and a internal validation set (n = 114) in a 7:3 ratio. In the training set, univariate analysis identified candidate indicators (P < 0.05). Multivariate logistic regression and Least Absolute Shrinkage and Selection Operator (LASSO) regression were subsequently used to select independent predictors and prevent overfitting. Three machine learning models-Random Forest, Gradient Boosting, and Support Vector Machine-were constructed. Model performance was evaluated using the Area Under the Receiver Operating Characteristic Curve (AUC), calibration curves, and Decision Curve Analysis. Interpretability was assessed via SHapley Additive exPlanations (SHAP) values. A visual nomogram was developed.

Baseline characteristics were comparable between the training and internal validation sets (P > 0.05). Five indicators were significantly associated with stunting, including infant weight Z-score, length Z-score, length growth velocity, diversity of complementary foods, and hemoglobin (Hb). Length growth velocity was the strongest predictor (OR=0.340, 95% CI: 0.211-0.548, P < 0.001). LASSO regression confirmed infant weight Z-score, length Z-score, length growth velocity, number of complementary food types, and Hb as the optimal variable combination. The Gradient Boosting model demonstrated superior performance, with an AUC of 0.861 (95% CI: 0.784-0.938) in the training set and 0.850 (95% CI: 0.699-1.000) in the internal validation set. Its calibration was excellent, and decision curve analysis indicated a higher net benefit across a wide risk threshold range. SHAP analysis identified infant weight Z-score as the most critical predictive variable. The nomogram provided a practical tool for quantitative risk assessment.

The developed nutritional scoring system and Gradient Boosting prediction algorithm exhibited robust performance in identifying infants at risk of stunting. This tool facilitates early quantitative risk assessment and supports targeted clinical interventions.