Incorporating machine learning and statistical methods to address maternal healthcare disparities in US: A systematic...

Maternal health disparities are recognized as a significant public health challenge, with pronounced disparities evident across racial, socioeconomic, and geographic dimensions. Although healthcare technologies have advanced, these disparities remain primarily unaddressed, indicating that enhanced analytical approaches are needed.

This review aims to evaluate the impact of machine learning (ML) and statistical methods on identifying and addressing maternal health disparities and to outline future research directions for enhancing these methodologies.

Following the PRISMA guidelines, the review of studies employing ML and statistical methods to analyze maternal health disparities within the United States was conducted. Publications between January 1, 2012, and February 2024 were systematically searched through PubMed, Web of Science, and ScienceDirect. Inclusion criteria targeted studies conducted within the U.S., peer-reviewed articles published during the period, research covering the postpartum period up to one year post-delivery, and studies incorporating both maternal and infant health data with a focus primarily on maternal outcomes.

A total of 147 studies met the inclusion criteria for this analysis. Among these, 129 (88 %) utilized statistical methods in health sciences to analyze correlations, treatment effects, and public health initiatives, thus providing vital, actionable insights for policy and clinical decisions. Meanwhile, 18 articles (12 %) applied ML techniques to explore complex, nonlinear relationships in data. The findings indicate that while ML and statistical methods offer valuable insights into the factors contributing to health disparities, there are limitations regarding dataset diversity and methodological precision. Most studies concentrate on racial and socioeconomic inequalities, with fewer addressing the geographical aspects of maternal health. This review emphasizes the necessity for broader dataset utilization and methodology improvements to enhance the findings' predictive accuracy and applicability.

ML and statistical methods show great potential to transform maternal healthcare by identifying and addressing disparities. Future research should focus on broadening dataset diversity, improving methodological precision, and enhancing interdisciplinary efforts.