Precision at its Core: Machine Learning-Infused Metabolomics Model for Preterm Birth Prediction in Human

Preterm Birth is a leading cause of neonatal morbidity and mortality, and it can also have long-term effects on the health and well-being of the child, including developmental disabilities, cognitive impairments, and chronic health conditions. This study aims to offer a prediction model based on a machine learning algorithm that predicts PTB using metabolomics data. Five separate models are used in the pipeline: Random Forest, K-Nearest Neighbors, Gradient Boosting, Decision Tree, and Logistic Regression. The models display different degrees of AUC, recall, F1 score, and ROC. The best performance was shown by the Gradient Boosting model (accuracy:0.97; AUC:0.98) which predicted PTB exceptionally well compared to other models. Calcium carbonate, Thymidine-5’-carboxylic acid, 2-Hydroxyfelbamate, (2R,3R)-3-chloro-6-hydromellein, and Fenothiocarbsulfoxide were the top 5 metabolites that contribute towards the pathogenesis of PTB. The interventions for the main metabolites of PTB in the early stages could reduce its risk.