Work

Maternal Nutritional Factors Enhance Birthweight Prediction: A Super Learner Ensemble Approach

Year: 2024

Type: article

Abstract: Background and Objectives: Birthweight (BW) is a widely used indicator of postnatal well-being. Low birthweight (LBW) is an indicator of neonatal morbidity and mortality risk. Timely and precise prediction of LBW is crucial for ensuring newborn health and well-being. Despite recent machine learning advancements in LBW classification based on physiological traits in the mother and ultrasound outcomes, maternal status in essential micronutrients for fetal development are yet to be fully exploited for BW prediction. The aim of this study is to investigate the contribution and impact of maternal nutritional features, including mid-pregnancy plasma Vitamin B12, folate, and iron status in the BW prediction.Methods: This study is based on data from pregnant women recruited from two university hospitals in the Tarragona province of Spain, encompassing a dataset of 729 mother-child dyads. The ensemble learning approach called super learner was utilized to predict neonate BW early. To evaluate the super learner model, ten cross-validations were applied, and different evaluation metrics were used: accuracy, precision, recall, F1-score, and Area under curve receiver operating characteristic (AUC-ROC). Each feature’s contribution and impact were also computed using the feature importance technique and partial dependency plot.Results: The model accurately predicts BW class by leveraging mid-pregnancy physiological, nutritional and lifestyle factors associated with LBW (≤2500 g) and preterm birth. The results illustrate the superiority of the super learner model, surpassing single-based classification ML models with an accuracy of 96.19% and an AUC of 0.96. This study highlights the significance of maternal vitamin B12 and plasma folate status during pregnancy as the most influential attributes in predicting BW.Conclusions: The studied maternal nutritional factors significantly enhanced the performance of BW prediction using the ensemble ML models, mainly based on super learner. The findings identified plasma folate and vitamin B12 as essential markers for health professionals to control LBW, neonate illness and mortality.

Source: Information

DOI: 10.3390/info15110714

Authors: Muhammad Mursil, Hatem A. Rashwan, Pere Cavallé-Busquets, Luis A. Santos-Calderón, Michelle M. Murphy, Domenec Puig

Institutions: Universitat Rovira i Virgili, Institut de Recerca Biomèdica Catalunya Sud, Hospital Universitari Sant Joan de Reus

Language: English

Topic: Artificial Intelligence in Healthcare

Subfield: Health Information Management

Field: Health Professions

Domain: Health Sciences

SDGs: Good health and well-being, Zero hunger

Open Access status: gold

Estimated APC paid: $1,819