Background and Aims: Low birth weight (LBW), known as the condition of a newborn weighing less than 2500 g, is a growing concern in the United States (US). Previous studies have identified some factors contributing to this issue. However, these studies have primarily focused on individual factors rather than examining the correlations between multiple factors. Additionally, much of the research has concentrated on maternal health, demographics, and socioeconomic factors, often overlooking potential influences from father-related factors such as paternal age, educational level, and ethnicity. Moreover, many of these studies have been limited to local populations, which may not represent the entire US population. To address these limitations, this study employs a machine learning approach to analyze the 2022 Centers for Disease Control and Prevention's National Natality Dataset, identifying the most significant factors contributing to LBW across the US.
Methods: Relevant variables such as anthropometric, socioeconomic, maternal, and paternal factors from the dataset were used to train logistic regression, random forest, extreme gradient boosting, and an attention mechanism model to predict normal and LBW newborns. The predictive models were interpretable through odds ratio, feature importance, and Shapley Additive Explanations (SHAP), revealing the factors contributing most to LBW.
Results: Our approach predicted normal birth weight and LBW newborns with an accuracy of 70%, 73%, 74%, and 59.5% for the logistic regression, random forest, extreme gradient boosting, and an attention mechanism model, respectively. The models indicated that both maternal and paternal factors influence birth weights. Among these factors, ethnicity, age, and education level were the most common among parents. Additionally, maternal characteristics, like height, BMI, pre-pregnancy weight, and weight gain during pregnancy, were significant in determining birth weights.
Conclusion: Our findings provide insights to support pregnancy guidelines. Specifically, parents should prioritize early and regular prenatal consultations, proper nutrition, and avoiding smoking. This should be a priority, particularly for vulnerable populations in which LBW is a high burden. Moreover, the significant influence of ethnicity and maternal height on LBW suggests the need to reconsider the 2500-gram cut-off established by the World Health Organization to account for the differences in parent height across ethnicities.