Development of Machine Learning Models for Early Prediction of Small- for-Gestational-Age Births Using Maternal Sociodemographic and Obstetric Data

Background: Small for gestational age (SGA) is a significant concern in obstetrics, with implications for stillbirth, neonatal mortality, and long-term health outcomes. The early detection of SGA is crucial for prevention and treatment, but current methods have limitations. This study aimed to develop a machine learning (ML) models-based algorithm to predict SGA using sociodemographic and obstetric features during the preconception period. Methods: This study obtained information from pregnant women when they first entered the obstetric clinic. A comprehensive analysis included maternal age determination together with body mass index measurements and values for gravida status and parity numbers along with assessments of previous birth weight and records of hypertension and/or diabetes mellitus (DM) diagnosed before pregnancy and preeclampsia and gestational DM. The study participants were divided into groups based on having delivered small for gestational age (SGA) live birth infants or being nulliparous or primiparous. Standard ML algorithms comprised five different structures were used to develop the predictive model. Results: The study examined 102 mothers while 27 and 26 nulliparous were with and without an SGA birth, respectively. 25 and 24 primiparous were with and without an SGA birth, respectively. The primary defining variables for predicting SGA index as per the analysis were weight and body mass index and height and previous birth weight. The Logistic Regression model showed the optimal performance for nulliparous individuals with 72.7% accuracy yet the extreme Gradient Boosting model excelled at predicting outcomes for primiparous individuals with 80% accuracy. Conclusion: An ML model constructed with basic maternal sociodemographic findings and obstetric history may serve as an early prediction tool for SGA during the pre-conception period, particularly in resource-constrained settings, although broader validation is required.